MindWorkAI/AI-Studio
MindWorkAI/AI-Studio
1
0
Fork 0
mirror of https://github.com/MindWorkAI/AI-Studio.git synced 2026-09-27 02:53:38 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Name the tokenizer a model uses, where its vendor does

Browse Source
This commit is contained in:
Thorsten Sommer 2026-09-12 19:59:56 +02:00
parent 2509fc09d1
commit 5326a0bf6c
Signed by untrusted user who does not match committer: tsommer
GPG Key ID: 371BBA77A02C0108
16 changed files with 521 additions and 285 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
app/MindWork AI Studio/Assistants/I18N/allTexts.lua
View File

@ -7063,6 +7063,9 @@ UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T3804472591"] = "Duplicate k
-- Override Model Capabilities -- Override Model Capabilities
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T3904244586"] = "Override Model Capabilities" UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T3904244586"] = "Override Model Capabilities"
-- The vendor of this model publishes no tokenizer file and counts through their API instead ({0}). AI Studio therefore estimates the token count with its built-in tokenizer.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T3965340739"] = "The vendor of this model publishes no tokenizer file and counts through their API instead ({0}). AI Studio therefore estimates the token count with its built-in tokenizer."
-- Currently, we cannot query the models for the selected provider and/or host. Therefore, please enter the model name manually. -- Currently, we cannot query the models for the selected provider and/or host. Therefore, please enter the model name manually.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4116737656"] = "Currently, we cannot query the models for the selected provider and/or host. Therefore, please enter the model name manually." UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4116737656"] = "Currently, we cannot query the models for the selected provider and/or host. Therefore, please enter the model name manually."
@ -7078,6 +7081,9 @@ UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4285779702"] = "Choose File
-- Video input -- Video input
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4289835208"] = "Video input" UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4289835208"] = "Video input"
-- This model uses OpenAI's {0} encoding, which does not come as a tokenizer.json file. AI Studio therefore estimates the token count with its built-in tokenizer.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T466506475"] = "This model uses OpenAI's {0} encoding, which does not come as a tokenizer.json file. AI Studio therefore estimates the token count with its built-in tokenizer."
-- We are currently unable to communicate with the provider to load models. Please try again later. -- We are currently unable to communicate with the provider to load models. Please try again later.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T504465522"] = "We are currently unable to communicate with the provider to load models. Please try again later." UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T504465522"] = "We are currently unable to communicate with the provider to load models. Please try again later."
@ -7099,6 +7105,9 @@ UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T900237532"] = "Provider"
-- Cancel -- Cancel
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T900713019"] = "Cancel" UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T900713019"] = "Cancel"
-- This model uses the tokenizer of {0}. Download its tokenizer.json file and select it below to count exactly instead of estimating.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T924854143"] = "This model uses the tokenizer of {0}. Download its tokenizer.json file and select it below to count exactly instead of estimating."
-- For better token estimates, you can configure a custom tokenizer for this provider. -- For better token estimates, you can configure a custom tokenizer for this provider.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T961454300"] = "For better token estimates, you can configure a custom tokenizer for this provider." UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T961454300"] = "For better token estimates, you can configure a custom tokenizer for this provider."

7
app/MindWork AI Studio/Dialogs/ProviderDialog.razor
View File

@ -251,6 +251,13 @@
<MudJustifiedText Typo="Typo.body1" Class="mt-4 mb-3"> <MudJustifiedText Typo="Typo.body1" Class="mt-4 mb-3">
@T("For better token estimates, you can configure a custom tokenizer for this provider.") @T("For better token estimates, you can configure a custom tokenizer for this provider.")
</MudJustifiedText> </MudJustifiedText>
var tokenizerHint = this.GetCurrentModelTokenizerLabel();
if (!string.IsNullOrEmpty(tokenizerHint))
{
<MudJustifiedText Typo="Typo.body2" Class="mb-3">
@tokenizerHint
</MudJustifiedText>
}
<PathDropZone IdPrefix="tokenizer" Disabled="@(() => this.IsEnterpriseConfiguration)" OnPathsDropped="@this.OnTokenizerPathsDropped"> <PathDropZone IdPrefix="tokenizer" Disabled="@(() => this.IsEnterpriseConfiguration)" OnPathsDropped="@this.OnTokenizerPathsDropped">
<MudStack Row="@true" Spacing="3" Class="mb-3" StretchItems="StretchItems.None" AlignItems="AlignItems.Center"> <MudStack Row="@true" Spacing="3" Class="mb-3" StretchItems="StretchItems.None" AlignItems="AlignItems.Center">
<MudTextField <MudTextField

29
app/MindWork AI Studio/Dialogs/ProviderDialog.razor.cs
View File

@ -778,6 +778,35 @@ public partial class ProviderDialog : MSGComponentBase, ISecretId
return "Unknown"; return "Unknown";
} }
/// <summary>
/// What the dialog says about the tokenizer of the current model, if anything.
/// </summary>
/// <remarks>
/// The field below this sentence takes a tokenizer.json file and nothing else, and until now it
/// said nothing about which file. That leaves two kinds of people stuck: the ones who could
/// download the right file and do not know its name, and the ones who spend an evening looking
/// for Anthropic's tokenizer file, which does not exist.
///
/// Empty for a model nobody stated a tokenizer for, which is most of them. An empty sentence
/// hides the whole line rather than claiming that nothing is known about it.
/// </remarks>
/// <returns>The sentence, or nothing.</returns>
private string GetCurrentModelTokenizerLabel()
{
var tokenizer = this.GetCurrentModelProfile().Tokenizer;
if (!tokenizer.IsKnown)
return string.Empty;
return tokenizer.Kind switch
{
TokenizerKind.HUGGING_FACE => string.Format(T("This model uses the tokenizer of {0}. Download its tokenizer.json file and select it below to count exactly instead of estimating."), tokenizer.Id),
TokenizerKind.TIKTOKEN => string.Format(T("This model uses OpenAI's {0} encoding, which does not come as a tokenizer.json file. AI Studio therefore estimates the token count with its built-in tokenizer."), tokenizer.Id),
TokenizerKind.PROVIDER_API => string.Format(T("The vendor of this model publishes no tokenizer file and counts through their API instead ({0}). AI Studio therefore estimates the token count with its built-in tokenizer."), tokenizer.Id),
_ => string.Empty,
};
}
private string GetCapabilityOverrideLabel(Capability capability) => capability switch private string GetCapabilityOverrideLabel(Capability capability) => capability switch
{ {
Capability.AUDIO_INPUT => T("Audio input"), Capability.AUDIO_INPUT => T("Audio input"),

6
app/MindWork AI Studio/Models/Anthropic/ClaudeFamily.cs
View File

@ -52,7 +52,8 @@ public sealed class ClaudeFamily : ModelFamily
/// <inheritdoc /> /// <inheritdoc />
public override IReadOnlyList<ModelSource> FurtherSources => public override IReadOnlyList<ModelSource> FurtherSources =>
[ [
new("https://platform.claude.com/docs/en/build-with-claude/vision", new DateOnly(2026, 9, 12), "The vision page gives the image limit as a rule rather than as a number per model: 100 images per request on the API for models with a 200k-token context window, 600 per request for all other models. The 20 it also names belongs to claude.ai, not to the API.") new("https://platform.claude.com/docs/en/build-with-claude/vision", new DateOnly(2026, 9, 12), "The vision page gives the image limit as a rule rather than as a number per model: 100 images per request on the API for models with a 200k-token context window, 600 per request for all other models. The 20 it also names belongs to claude.ai, not to the API."),
new("https://platform.claude.com/docs/en/build-with-claude/token-counting", new DateOnly(2026, 9, 12), "Anthropic publishes no tokenizer file at all; they count through the /v1/messages/count_tokens endpoint instead. The same page warns that Claude 4.7 and later use a newer tokenizer, on which the same text counts roughly 30 percent higher -- so AI Studio's built-in estimate is further off for those models than for the older ones.")
]; ];
/// <inheritdoc /> /// <inheritdoc />
@ -84,7 +85,8 @@ public sealed class ClaudeFamily : ModelFamily
builder.Rule("claude").AsSegment() builder.Rule("claude").AsSegment()
.Capabilities(TEXT_INPUT | MULTIPLE_IMAGE_INPUT | TEXT_OUTPUT | FUNCTION_CALLING) .Capabilities(TEXT_INPUT | MULTIPLE_IMAGE_INPUT | TEXT_OUTPUT | FUNCTION_CALLING)
.Apis(CHAT_COMPLETION_API) .Apis(CHAT_COMPLETION_API)
.ContextWindow(STANDARD_WINDOW); .ContextWindow(STANDARD_WINDOW)
.Tokenizer(TokenizerKind.PROVIDER_API, "/v1/messages/count_tokens");
// //
// The 3.x models say nothing beyond the shape above, so nothing is written for them: the // The 3.x models say nothing beyond the shape above, so nothing is written for them: the

6
app/MindWork AI Studio/Models/Google/GeminiFamily.cs
View File

@ -29,7 +29,8 @@ public sealed class GeminiFamily : ModelFamily
/// <inheritdoc /> /// <inheritdoc />
public override IReadOnlyList<ModelSource> FurtherSources => public override IReadOnlyList<ModelSource> FurtherSources =>
[ [
new("https://ai.google.dev/gemini-api/docs/image-understanding", new DateOnly(2026, 9, 12), "States one number for the whole family: \"Gemini models support a maximum of 3,600 image files per request.\" The 20 MB it also names is a limit on the request body rather than on the number of images.") new("https://ai.google.dev/gemini-api/docs/image-understanding", new DateOnly(2026, 9, 12), "States one number for the whole family: \"Gemini models support a maximum of 3,600 image files per request.\" The 20 MB it also names is a limit on the request body rather than on the number of images."),
new("https://ai.google.dev/gemini-api/docs/tokens", new DateOnly(2026, 9, 12), "Google publishes no tokenizer file for Gemini. Counting happens through the countTokens method of the API, which returns the number of tokens of the input alone.")
]; ];
/// <inheritdoc /> /// <inheritdoc />
@ -44,7 +45,8 @@ public sealed class GeminiFamily : ModelFamily
builder.Rule("gemini").AsSegment() builder.Rule("gemini").AsSegment()
.Capabilities(TEXT_INPUT | MULTIPLE_IMAGE_INPUT | AUDIO_INPUT | SPEECH_INPUT | VIDEO_INPUT | TEXT_OUTPUT | FUNCTION_CALLING) .Capabilities(TEXT_INPUT | MULTIPLE_IMAGE_INPUT | AUDIO_INPUT | SPEECH_INPUT | VIDEO_INPUT | TEXT_OUTPUT | FUNCTION_CALLING)
.Apis(CHAT_COMPLETION_API) .Apis(CHAT_COMPLETION_API)
.Images(maxPerRequest: 3_600); .Images(maxPerRequest: 3_600)
.Tokenizer(TokenizerKind.PROVIDER_API, "countTokens");
// The one Gemini which only ever read text and images: // The one Gemini which only ever read text and images:
builder.Rule("gemini-1.0-pro-vision").AsPrefix() builder.Rule("gemini-1.0-pro-vision").AsPrefix()

12
app/MindWork AI Studio/Models/OpenAI/Gpt35Family.cs
View File

@ -13,12 +13,19 @@ public sealed class Gpt35Family : ModelFamily
/// <inheritdoc /> /// <inheritdoc />
public override ModelSource Source => new("https://platform.openai.com/docs/models", new DateOnly(2026, 9, 11), "Ported unchanged from the rules in ProviderExtensions.OpenAI.cs: text in, text out, no tools and no images."); public override ModelSource Source => new("https://platform.openai.com/docs/models", new DateOnly(2026, 9, 11), "Ported unchanged from the rules in ProviderExtensions.OpenAI.cs: text in, text out, no tools and no images.");
/// <inheritdoc />
public override IReadOnlyList<ModelSource> FurtherSources =>
[
new("https://github.com/openai/tiktoken/blob/main/tiktoken/model.py", new DateOnly(2026, 9, 12), "OpenAI's own mapping from model names to encodings. It maps \"gpt-3.5\", \"gpt-3.5-turbo\" and the prefix \"gpt-3.5-turbo-\" to cl100k_base.")
];
/// <inheritdoc /> /// <inheritdoc />
protected override void Declare(ModelFamilyBuilder builder) protected override void Declare(ModelFamilyBuilder builder)
{ {
builder.Rule("gpt-3.5").AsPrefix() builder.Rule("gpt-3.5").AsPrefix()
.Capabilities(TEXT_INPUT | TEXT_OUTPUT) .Capabilities(TEXT_INPUT | TEXT_OUTPUT)
.Apis(CHAT_COMPLETION_API); .Apis(CHAT_COMPLETION_API)
.Tokenizer(TokenizerKind.TIKTOKEN, "cl100k_base");
// //
// The odd one out, and kept odd on purpose: the previous rules put this one model on the // The odd one out, and kept odd on purpose: the previous rules put this one model on the
@ -28,6 +35,7 @@ public sealed class Gpt35Family : ModelFamily
// //
builder.Rule("gpt-3.5-turbo").AsExact() builder.Rule("gpt-3.5-turbo").AsExact()
.Capabilities(TEXT_INPUT | TEXT_OUTPUT) .Capabilities(TEXT_INPUT | TEXT_OUTPUT)
.Apis(RESPONSES_API); .Apis(RESPONSES_API)
.Tokenizer(TokenizerKind.TIKTOKEN, "cl100k_base");
} }
} }

9
app/MindWork AI Studio/Models/OpenAI/Gpt4Family.cs
View File

@ -18,13 +18,20 @@ public sealed class Gpt4Family : ModelFamily
/// <inheritdoc /> /// <inheritdoc />
public override ModelSource Source => new("https://developers.openai.com/api/docs/models/gpt-4-turbo", new DateOnly(2026, 9, 12), "Capabilities ported unchanged from ProviderExtensions.OpenAI.cs: GPT-4 is text only, Turbo adds images and tool calling. The windows are the documented 8,192 tokens of GPT-4 and the 128,000 Turbo raised it to."); public override ModelSource Source => new("https://developers.openai.com/api/docs/models/gpt-4-turbo", new DateOnly(2026, 9, 12), "Capabilities ported unchanged from ProviderExtensions.OpenAI.cs: GPT-4 is text only, Turbo adds images and tool calling. The windows are the documented 8,192 tokens of GPT-4 and the 128,000 Turbo raised it to.");
/// <inheritdoc />
public override IReadOnlyList<ModelSource> FurtherSources =>
[
new("https://github.com/openai/tiktoken/blob/main/tiktoken/model.py", new DateOnly(2026, 9, 12), "OpenAI's own mapping from model names to encodings. It maps \"gpt-4\" and the prefix \"gpt-4-\" to cl100k_base, so Turbo uses it too -- the newer o200k_base begins with the 4o line, which is a family of its own here.")
];
/// <inheritdoc /> /// <inheritdoc />
protected override void Declare(ModelFamilyBuilder builder) protected override void Declare(ModelFamilyBuilder builder)
{ {
builder.Rule("gpt-4").AsPrefix() builder.Rule("gpt-4").AsPrefix()
.Capabilities(TEXT_INPUT | TEXT_OUTPUT) .Capabilities(TEXT_INPUT | TEXT_OUTPUT)
.Apis(RESPONSES_API) .Apis(RESPONSES_API)
.ContextWindow(8_192); .ContextWindow(8_192)
.Tokenizer(TokenizerKind.TIKTOKEN, "cl100k_base");
builder.Rule("gpt-4-turbo").AsPrefix().Inherits() builder.Rule("gpt-4-turbo").AsPrefix().Inherits()
.Capabilities(MULTIPLE_IMAGE_INPUT | FUNCTION_CALLING) .Capabilities(MULTIPLE_IMAGE_INPUT | FUNCTION_CALLING)

15
app/MindWork AI Studio/Models/OpenAI/Gpt4oFamily.cs
View File

@ -20,13 +20,20 @@ public sealed class Gpt4oFamily : ModelFamily
/// <inheritdoc /> /// <inheritdoc />
public override ModelSource Source => new("https://developers.openai.com/api/docs/models/gpt-4o", new DateOnly(2026, 9, 12), "The answer the previous rules gave these models through their fallback: images, tool calling, and web search on the Responses API. The model page states a 128,000 token window, which the minis and the search previews share."); public override ModelSource Source => new("https://developers.openai.com/api/docs/models/gpt-4o", new DateOnly(2026, 9, 12), "The answer the previous rules gave these models through their fallback: images, tool calling, and web search on the Responses API. The model page states a 128,000 token window, which the minis and the search previews share.");
/// <inheritdoc />
public override IReadOnlyList<ModelSource> FurtherSources =>
[
new("https://github.com/openai/tiktoken/blob/main/tiktoken/model.py", new DateOnly(2026, 9, 12), "OpenAI's own mapping from model names to encodings. It maps the prefix \"gpt-4o-\" to o200k_base, which covers the minis and the search previews as well.")
];
/// <inheritdoc /> /// <inheritdoc />
protected override void Declare(ModelFamilyBuilder builder) protected override void Declare(ModelFamilyBuilder builder)
{ {
builder.Rule("gpt-4o").AsPrefix() builder.Rule("gpt-4o").AsPrefix()
.Capabilities(TEXT_INPUT | MULTIPLE_IMAGE_INPUT | TEXT_OUTPUT | FUNCTION_CALLING | WEB_SEARCH) .Capabilities(TEXT_INPUT | MULTIPLE_IMAGE_INPUT | TEXT_OUTPUT | FUNCTION_CALLING | WEB_SEARCH)
.Apis(RESPONSES_API) .Apis(RESPONSES_API)
.ContextWindow(128_000); .ContextWindow(128_000)
.Tokenizer(TokenizerKind.TIKTOKEN, "o200k_base");
// //
// The search previews are the same generation and almost nothing like it: they search the // The search previews are the same generation and almost nothing like it: they search the
@ -37,11 +44,13 @@ public sealed class Gpt4oFamily : ModelFamily
builder.Rule("gpt-4o-search-preview").AsExact() builder.Rule("gpt-4o-search-preview").AsExact()
.Capabilities(TEXT_INPUT | TEXT_OUTPUT | WEB_SEARCH) .Capabilities(TEXT_INPUT | TEXT_OUTPUT | WEB_SEARCH)
.Apis(CHAT_COMPLETION_API) .Apis(CHAT_COMPLETION_API)
.ContextWindow(128_000); .ContextWindow(128_000)
.Tokenizer(TokenizerKind.TIKTOKEN, "o200k_base");
builder.Rule("gpt-4o-mini-search-preview").AsExact() builder.Rule("gpt-4o-mini-search-preview").AsExact()
.Capabilities(TEXT_INPUT | TEXT_OUTPUT | WEB_SEARCH) .Capabilities(TEXT_INPUT | TEXT_OUTPUT | WEB_SEARCH)
.Apis(CHAT_COMPLETION_API) .Apis(CHAT_COMPLETION_API)
.ContextWindow(128_000); .ContextWindow(128_000)
.Tokenizer(TokenizerKind.TIKTOKEN, "o200k_base");
} }
} }

9
app/MindWork AI Studio/Models/OpenAI/Gpt5Family.cs
View File

@ -26,6 +26,12 @@ public sealed class Gpt5Family : ModelFamily
/// <inheritdoc /> /// <inheritdoc />
public override ModelSource Source => new("https://developers.openai.com/api/docs/models", new DateOnly(2026, 9, 12), "Capabilities ported unchanged from ProviderExtensions.OpenAI.cs, one rule per generation, except that the chat alias no longer inherits the reasoning it is named for not having. Context windows read per generation from the model pages below that URL."); public override ModelSource Source => new("https://developers.openai.com/api/docs/models", new DateOnly(2026, 9, 12), "Capabilities ported unchanged from ProviderExtensions.OpenAI.cs, one rule per generation, except that the chat alias no longer inherits the reasoning it is named for not having. Context windows read per generation from the model pages below that URL.");
/// <inheritdoc />
public override IReadOnlyList<ModelSource> FurtherSources =>
[
new("https://github.com/openai/tiktoken/blob/main/tiktoken/model.py", new DateOnly(2026, 9, 12), "OpenAI's own mapping from model names to encodings. It maps the prefix \"gpt-5\" to o200k_base, which covers every model of this line.")
];
/// <inheritdoc /> /// <inheritdoc />
protected override void Declare(ModelFamilyBuilder builder) protected override void Declare(ModelFamilyBuilder builder)
{ {
@ -38,7 +44,8 @@ public sealed class Gpt5Family : ModelFamily
.Capabilities(TEXT_INPUT | MULTIPLE_IMAGE_INPUT | TEXT_OUTPUT | FUNCTION_CALLING | WEB_SEARCH) .Capabilities(TEXT_INPUT | MULTIPLE_IMAGE_INPUT | TEXT_OUTPUT | FUNCTION_CALLING | WEB_SEARCH)
.Apis(RESPONSES_API) .Apis(RESPONSES_API)
.Reasoning(ReasoningSupport.ALWAYS) .Reasoning(ReasoningSupport.ALWAYS)
.ContextWindow(400_000); .ContextWindow(400_000)
.Tokenizer(TokenizerKind.TIKTOKEN, "o200k_base");
// //
// The alias for the model of this generation which does not reason. The previous rules had // The alias for the model of this generation which does not reason. The previous rules had

18
app/MindWork AI Studio/Models/OpenAI/OSeriesFamily.cs
View File

@ -22,6 +22,12 @@ public sealed class OSeriesFamily : ModelFamily
/// <inheritdoc /> /// <inheritdoc />
public override ModelSource Source => new("https://developers.openai.com/api/docs/models", new DateOnly(2026, 9, 12), "Capabilities ported unchanged from ProviderExtensions.OpenAI.cs, one rule per generation and one per mini. The o1 and o3 pages state 200,000 tokens; the two cut-down minis have no page of their own, so no window is stated for them."); public override ModelSource Source => new("https://developers.openai.com/api/docs/models", new DateOnly(2026, 9, 12), "Capabilities ported unchanged from ProviderExtensions.OpenAI.cs, one rule per generation and one per mini. The o1 and o3 pages state 200,000 tokens; the two cut-down minis have no page of their own, so no window is stated for them.");
/// <inheritdoc />
public override IReadOnlyList<ModelSource> FurtherSources =>
[
new("https://github.com/openai/tiktoken/blob/main/tiktoken/model.py", new DateOnly(2026, 9, 12), "OpenAI's own mapping from model names to encodings. It maps \"o1\", \"o3\", \"o4-mini\" and the prefixes \"o1-\", \"o3-\" and \"o4-mini-\" to o200k_base, so the whole series shares one encoding.")
];
/// <inheritdoc /> /// <inheritdoc />
protected override void Declare(ModelFamilyBuilder builder) protected override void Declare(ModelFamilyBuilder builder)
{ {
@ -29,23 +35,27 @@ public sealed class OSeriesFamily : ModelFamily
.Capabilities(TEXT_INPUT | MULTIPLE_IMAGE_INPUT | TEXT_OUTPUT | FUNCTION_CALLING) .Capabilities(TEXT_INPUT | MULTIPLE_IMAGE_INPUT | TEXT_OUTPUT | FUNCTION_CALLING)
.Apis(RESPONSES_API) .Apis(RESPONSES_API)
.Reasoning(ReasoningSupport.ALWAYS) .Reasoning(ReasoningSupport.ALWAYS)
.ContextWindow(200_000); .ContextWindow(200_000)
.Tokenizer(TokenizerKind.TIKTOKEN, "o200k_base");
builder.Rule("o1-mini").AsPrefix() builder.Rule("o1-mini").AsPrefix()
.Capabilities(TEXT_INPUT | TEXT_OUTPUT) .Capabilities(TEXT_INPUT | TEXT_OUTPUT)
.Apis(CHAT_COMPLETION_API) .Apis(CHAT_COMPLETION_API)
.Reasoning(ReasoningSupport.ALWAYS); .Reasoning(ReasoningSupport.ALWAYS)
.Tokenizer(TokenizerKind.TIKTOKEN, "o200k_base");
builder.Rule("o3").AsPrefix() builder.Rule("o3").AsPrefix()
.Capabilities(TEXT_INPUT | MULTIPLE_IMAGE_INPUT | TEXT_OUTPUT | FUNCTION_CALLING | WEB_SEARCH) .Capabilities(TEXT_INPUT | MULTIPLE_IMAGE_INPUT | TEXT_OUTPUT | FUNCTION_CALLING | WEB_SEARCH)
.Apis(RESPONSES_API) .Apis(RESPONSES_API)
.Reasoning(ReasoningSupport.ALWAYS) .Reasoning(ReasoningSupport.ALWAYS)
.ContextWindow(200_000); .ContextWindow(200_000)
.Tokenizer(TokenizerKind.TIKTOKEN, "o200k_base");
builder.Rule("o3-mini").AsPrefix() builder.Rule("o3-mini").AsPrefix()
.Capabilities(TEXT_INPUT | TEXT_OUTPUT | FUNCTION_CALLING) .Capabilities(TEXT_INPUT | TEXT_OUTPUT | FUNCTION_CALLING)
.Apis(RESPONSES_API) .Apis(RESPONSES_API)
.Reasoning(ReasoningSupport.ALWAYS); .Reasoning(ReasoningSupport.ALWAYS)
.Tokenizer(TokenizerKind.TIKTOKEN, "o200k_base");
// The one mini which is not cut down: it is the o3 generation under another number. // The one mini which is not cut down: it is the o3 generation under another number.
builder.Rule("o4-mini").AsPrefix().InheritsFrom("o3"); builder.Rule("o4-mini").AsPrefix().InheritsFrom("o3");

9
app/MindWork AI Studio/Plugins/languages/de-de-43065dbc-78d0-45b7-92be-f14c2926e2dc/plugin.lua
View File

@ -7065,6 +7065,9 @@ UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T3804472591"] = "Doppelter S
-- Override Model Capabilities -- Override Model Capabilities
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T3904244586"] = "Modellfähigkeiten überschreiben" UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T3904244586"] = "Modellfähigkeiten überschreiben"
-- The vendor of this model publishes no tokenizer file and counts through their API instead ({0}). AI Studio therefore estimates the token count with its built-in tokenizer.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T3965340739"] = "Der Anbieter dieses Modells veröffentlicht keine Tokenizer-Datei und zählt die Tokens stattdessen über seine API ({0}). AI Studio schätzt die Anzahl der Tokens daher mit seinem integrierten Tokenizer."
-- Currently, we cannot query the models for the selected provider and/or host. Therefore, please enter the model name manually. -- Currently, we cannot query the models for the selected provider and/or host. Therefore, please enter the model name manually.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4116737656"] = "Derzeit können wir die Modelle für den ausgewählten Anbieter und/oder Host nicht abfragen. Bitte geben Sie daher den Modellnamen manuell ein." UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4116737656"] = "Derzeit können wir die Modelle für den ausgewählten Anbieter und/oder Host nicht abfragen. Bitte geben Sie daher den Modellnamen manuell ein."
@ -7080,6 +7083,9 @@ UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4285779702"] = "Datei ausw
-- Video input -- Video input
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4289835208"] = "Videoeingabe" UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4289835208"] = "Videoeingabe"
-- This model uses OpenAI's {0} encoding, which does not come as a tokenizer.json file. AI Studio therefore estimates the token count with its built-in tokenizer.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T466506475"] = "Dieses Modell verwendet die {0}-Kodierung von OpenAI, die nicht als tokenizer.json-Datei vorliegt. AI Studio schätzt die Anzahl der Token daher mit seinem integrierten Tokenizer."
-- We are currently unable to communicate with the provider to load models. Please try again later. -- We are currently unable to communicate with the provider to load models. Please try again later.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T504465522"] = "Wir können derzeit nicht mit dem Anbieter kommunizieren, um Modelle zu laden. Bitte versuchen Sie es später erneut." UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T504465522"] = "Wir können derzeit nicht mit dem Anbieter kommunizieren, um Modelle zu laden. Bitte versuchen Sie es später erneut."
@ -7101,6 +7107,9 @@ UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T900237532"] = "Anbieter"
-- Cancel -- Cancel
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T900713019"] = "Abbrechen" UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T900713019"] = "Abbrechen"
-- This model uses the tokenizer of {0}. Download its tokenizer.json file and select it below to count exactly instead of estimating.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T924854143"] = "Dieses Modell verwendet den Tokenizer von {0}. Laden Sie die Datei „tokenizer.json“ herunter und wählen Sie sie unten aus, um exakt statt nur geschätzt zu zählen."
-- For better token estimates, you can configure a custom tokenizer for this provider. -- For better token estimates, you can configure a custom tokenizer for this provider.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T961454300"] = "Für genauere Token-Schätzungen können Sie einen benutzerdefinierten Tokenizer für diesen Anbieter konfigurieren." UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T961454300"] = "Für genauere Token-Schätzungen können Sie einen benutzerdefinierten Tokenizer für diesen Anbieter konfigurieren."

9
app/MindWork AI Studio/Plugins/languages/en-us-97dfb1ba-50c4-4440-8dfa-6575daf543c8/plugin.lua
View File

@ -7065,6 +7065,9 @@ UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T3804472591"] = "Duplicate k
-- Override Model Capabilities -- Override Model Capabilities
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T3904244586"] = "Override Model Capabilities" UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T3904244586"] = "Override Model Capabilities"
-- The vendor of this model publishes no tokenizer file and counts through their API instead ({0}). AI Studio therefore estimates the token count with its built-in tokenizer.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T3965340739"] = "The vendor of this model publishes no tokenizer file and counts through their API instead ({0}). AI Studio therefore estimates the token count with its built-in tokenizer."
-- Currently, we cannot query the models for the selected provider and/or host. Therefore, please enter the model name manually. -- Currently, we cannot query the models for the selected provider and/or host. Therefore, please enter the model name manually.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4116737656"] = "Currently, we cannot query the models for the selected provider and/or host. Therefore, please enter the model name manually." UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4116737656"] = "Currently, we cannot query the models for the selected provider and/or host. Therefore, please enter the model name manually."
@ -7080,6 +7083,9 @@ UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4285779702"] = "Choose File
-- Video input -- Video input
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4289835208"] = "Video input" UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T4289835208"] = "Video input"
-- This model uses OpenAI's {0} encoding, which does not come as a tokenizer.json file. AI Studio therefore estimates the token count with its built-in tokenizer.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T466506475"] = "This model uses OpenAI's {0} encoding, which does not come as a tokenizer.json file. AI Studio therefore estimates the token count with its built-in tokenizer."
-- We are currently unable to communicate with the provider to load models. Please try again later. -- We are currently unable to communicate with the provider to load models. Please try again later.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T504465522"] = "We are currently unable to communicate with the provider to load models. Please try again later." UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T504465522"] = "We are currently unable to communicate with the provider to load models. Please try again later."
@ -7101,6 +7107,9 @@ UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T900237532"] = "Provider"
-- Cancel -- Cancel
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T900713019"] = "Cancel" UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T900713019"] = "Cancel"
-- This model uses the tokenizer of {0}. Download its tokenizer.json file and select it below to count exactly instead of estimating.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T924854143"] = "This model uses the tokenizer of {0}. Download its tokenizer.json file and select it below to count exactly instead of estimating."
-- For better token estimates, you can configure a custom tokenizer for this provider. -- For better token estimates, you can configure a custom tokenizer for this provider.
UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T961454300"] = "For better token estimates, you can configure a custom tokenizer for this provider." UI_TEXT_CONTENT["AISTUDIO::DIALOGS::PROVIDERDIALOG::T961454300"] = "For better token estimates, you can configure a custom tokenizer for this provider."

8
app/Tests/Models/CapabilityCharacterizationTests.cs
View File

@ -27,11 +27,11 @@ public sealed class CapabilityCharacterizationTests
/// How many columns a snapshot line carries after the model ID. /// How many columns a snapshot line carries after the model ID.
/// </summary> /// </summary>
/// <remarks> /// <remarks>
/// The capabilities, the kind, the context window, and the image limit. Adding a column to the /// The capabilities, the kind, the context window, the image limit, and the tokenizer. Adding a
/// snapshot means raising this, and forgetting to would split a line inside its last column /// column to the snapshot means raising this, and forgetting to would split a line inside its
/// instead of in front of it -- which makes every model look changed at once. /// last column instead of in front of it -- which makes every model look changed at once.
/// </remarks> /// </remarks>
private const int TRAILING_COLUMNS = 4; private const int TRAILING_COLUMNS = 5;
[Test] [Test]
public void TheCorpusStillGetsTheAnswersTheSnapshotRecorded() public void TheCorpusStillGetsTheAnswersTheSnapshotRecorded()

30
app/Tests/Models/Corpus/CapabilitySnapshot.cs
View File

@ -47,6 +47,16 @@ public static class CapabilitySnapshot
/// </remarks> /// </remarks>
private const string NO_IMAGE_LIMIT = "(unknown)"; private const string NO_IMAGE_LIMIT = "(unknown)";
/// <summary>
/// What a model nobody named a tokenizer for is written as.
/// </summary>
/// <remarks>
/// Which is what the app already does for all of them: it counts with the tokenizer it ships
/// and says that the number is an estimate. Naming one changes nothing about the counting yet;
/// it tells a person which file to look for, and for two vendors that there is none.
/// </remarks>
private const string NO_TOKENIZER = "(unknown)";
private const string HEADER = private const string HEADER =
""" """
# What the rules answer, for every model of the corpus. # What the rules answer, for every model of the corpus.
@ -55,9 +65,9 @@ public static class CapabilitySnapshot
# the diff. Every line of it is a statement about a model which somebody has to agree with. # the diff. Every line of it is a statement about a model which somebody has to agree with.
# #
# Columns are provider, model ID as the provider reports it, the capabilities sorted by # Columns are provider, model ID as the provider reports it, the capabilities sorted by
# name, what the model is made for, its context window in tokens, and how many images it # name, what the model is made for, its context window in tokens, how many images it
# accepts. The ID stands here unchanged, so a line may well carry leading or trailing # accepts, and which tokenizer it uses. The ID stands here unchanged, so a line may well
# spaces. # carry leading or trailing spaces.
# #
# A window or an image limit written as "(unknown)" is one nobody has stated a source for. # A window or an image limit written as "(unknown)" is one nobody has stated a source for.
# That is a gap, not a claim: the app then shows a person how many tokens their conversation # That is a gap, not a claim: the app then shows a person how many tokens their conversation
@ -124,9 +134,21 @@ public static class CapabilitySnapshot
private static string Line(CorpusEntry entry) private static string Line(CorpusEntry entry)
{ {
var profile = entry.Provider.GetModelProfile(new Model(entry.ModelId, null)); var profile = entry.Provider.GetModelProfile(new Model(entry.ModelId, null));
return $"{entry.Provider} | {entry.ModelId} | {Describe(RebuiltRules.AsCapabilities(profile))} | {profile.Kind} | {Describe(profile.Context)} | {Describe(profile.Images)}"; return $"{entry.Provider} | {entry.ModelId} | {Describe(RebuiltRules.AsCapabilities(profile))} | {profile.Kind} | {Describe(profile.Context)} | {Describe(profile.Images)} | {Describe(profile.Tokenizer)}";
} }
/// <summary>
/// Writes a tokenizer reference the way a snapshot line does.
/// </summary>
/// <remarks>
/// The kind travels with the name, because the name alone would be a riddle: "o200k_base" is
/// not a repository somebody can open, and "/v1/messages/count_tokens" is not a file somebody
/// can download. What sort of thing it is decides what a person can do with it.
/// </remarks>
/// <param name="tokenizer">The reference to write.</param>
/// <returns>The reference, or a marker when nobody named one.</returns>
public static string Describe(TokenizerRef tokenizer) => tokenizer.IsKnown ? $"{tokenizer.Kind} {tokenizer.Id}" : NO_TOKENIZER;
/// <summary> /// <summary>
/// Writes an image limit the way a snapshot line does. /// Writes an image limit the way a snapshot line does.
/// </summary> /// </summary>

524
app/Tests/Models/Corpus/CapabilitySnapshot.txt
View File

@ -4,9 +4,9 @@
# the diff. Every line of it is a statement about a model which somebody has to agree with. # the diff. Every line of it is a statement about a model which somebody has to agree with.
# #
# Columns are provider, model ID as the provider reports it, the capabilities sorted by # Columns are provider, model ID as the provider reports it, the capabilities sorted by
# name, what the model is made for, its context window in tokens, and how many images it # name, what the model is made for, its context window in tokens, how many images it
# accepts. The ID stands here unchanged, so a line may well carry leading or trailing # accepts, and which tokenizer it uses. The ID stands here unchanged, so a line may well
# spaces. # carry leading or trailing spaces.
# #
# A window or an image limit written as "(unknown)" is one nobody has stated a source for. # A window or an image limit written as "(unknown)" is one nobody has stated a source for.
# That is a gap, not a claim: the app then shows a person how many tokens their conversation # That is a gap, not a claim: the app then shows a person how many tokens their conversation
@ -16,262 +16,262 @@
# Models the audit found a wrong answer for are not in here. They live in # Models the audit found a wrong answer for are not in here. They live in
# ExpectedChanges.cs, together with the answer they have to arrive at. # ExpectedChanges.cs, together with the answer they have to arrive at.
# #
ALIBABA_CLOUD | qvq-max | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qvq-max | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen-max-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen-max-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen-mt-plus | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen-mt-plus | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen-plus-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen-plus-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen-turbo-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen-turbo-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen-vl-max | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen-vl-max | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen2.5-14b-instruct-1m | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen2.5-14b-instruct-1m | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen2.5-72b-instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen2.5-72b-instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen2.5-omni-7b | AUDIO_INPUT, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, SPEECH_OUTPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen2.5-omni-7b | AUDIO_INPUT, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, SPEECH_OUTPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen2.5-vl-72b-instruct | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen2.5-vl-72b-instruct | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen3-235b-a22b | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen3-235b-a22b | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen3-omni-flash | AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, SPEECH_OUTPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen3-omni-flash | AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, SPEECH_OUTPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen3-vl-plus | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen3-vl-plus | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen3.5-plus | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen3.5-plus | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen3.6-max | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen3.6-max | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen3.7-max | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen3.7-max | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen3.7-max-2026-05-17 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen3.7-max-2026-05-17 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen3.7-max-2026-06-08 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen3.7-max-2026-06-08 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen3.7-max-preview | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen3.7-max-preview | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen3.8-27b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen3.8-27b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen3.8-flash | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwen3.8-flash | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwen3.8-max | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1000000 | (unknown) ALIBABA_CLOUD | qwen3.8-max | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1000000 | (unknown) | (unknown)
ALIBABA_CLOUD | qwq-32b | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwq-32b | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ALIBABA_CLOUD | qwq-plus | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) ALIBABA_CLOUD | qwq-plus | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
ANTHROPIC | claude-3-5-haiku-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request ANTHROPIC | claude-3-5-haiku-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request | PROVIDER_API /v1/messages/count_tokens
ANTHROPIC | claude-3-5-sonnet-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request ANTHROPIC | claude-3-5-sonnet-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request | PROVIDER_API /v1/messages/count_tokens
ANTHROPIC | claude-3-7-sonnet-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request ANTHROPIC | claude-3-7-sonnet-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request | PROVIDER_API /v1/messages/count_tokens
ANTHROPIC | claude-3-opus-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request ANTHROPIC | claude-3-opus-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request | PROVIDER_API /v1/messages/count_tokens
ANTHROPIC | claude-7-sonnet | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request ANTHROPIC | claude-7-sonnet | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request | PROVIDER_API /v1/messages/count_tokens
ANTHROPIC | claude-fable-5-1 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request ANTHROPIC | claude-fable-5-1 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request | PROVIDER_API /v1/messages/count_tokens
ANTHROPIC | claude-haiku-4-5-20251001 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request ANTHROPIC | claude-haiku-4-5-20251001 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request | PROVIDER_API /v1/messages/count_tokens
ANTHROPIC | claude-mythos-5 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request ANTHROPIC | claude-mythos-5 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request | PROVIDER_API /v1/messages/count_tokens
ANTHROPIC | claude-opus-4-0 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request ANTHROPIC | claude-opus-4-0 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request | PROVIDER_API /v1/messages/count_tokens
ANTHROPIC | claude-opus-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request ANTHROPIC | claude-opus-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request | PROVIDER_API /v1/messages/count_tokens
ANTHROPIC | claude-sonnet-4-0 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request ANTHROPIC | claude-sonnet-4-0 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | 100 per request | PROVIDER_API /v1/messages/count_tokens
ANTHROPIC | claude-sonnet-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request ANTHROPIC | claude-sonnet-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request | PROVIDER_API /v1/messages/count_tokens
DEEP_SEEK | deepseek-chat | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) DEEP_SEEK | deepseek-chat | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
DEEP_SEEK | deepseek-reasoner | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) DEEP_SEEK | deepseek-reasoner | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
DEEP_SEEK | deepseek-v3.2-exp | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) DEEP_SEEK | deepseek-v3.2-exp | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
DEEP_SEEK | deepseek-v4 | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | (unknown) DEEP_SEEK | deepseek-v4 | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | (unknown) | (unknown)
DEEP_SEEK | deepseek-v4-vision | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | (unknown) DEEP_SEEK | deepseek-v4-vision | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | (unknown) | (unknown)
FIREWORKS | accounts/fireworks/models/deepseek-v3 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) FIREWORKS | accounts/fireworks/models/deepseek-v3 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
FIREWORKS | accounts/fireworks/models/llama-v3p1-405b-instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) FIREWORKS | accounts/fireworks/models/llama-v3p1-405b-instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) | (unknown)
FIREWORKS | accounts/fireworks/models/qwen3-235b-a22b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) FIREWORKS | accounts/fireworks/models/qwen3-235b-a22b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
GOOGLE | gemini-1.0-pro-vision | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) GOOGLE | gemini-1.0-pro-vision | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
GOOGLE | gemini-2.0-flash | AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | 3600 per request GOOGLE | gemini-2.0-flash | AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | 3600 per request | PROVIDER_API countTokens
GOOGLE | gemini-2.0-flash-live-001 | AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, SPEECH_INPUT, SPEECH_OUTPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) GOOGLE | gemini-2.0-flash-live-001 | AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, SPEECH_INPUT, SPEECH_OUTPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) | (unknown)
GOOGLE | gemini-2.5-flash | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1048576 | 3600 per request GOOGLE | gemini-2.5-flash | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1048576 | 3600 per request | PROVIDER_API countTokens
GOOGLE | gemini-2.5-flash-image | CHAT_COMPLETION_API, IMAGE_OUTPUT, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | IMAGE_GENERATION | (unknown) | (unknown) GOOGLE | gemini-2.5-flash-image | CHAT_COMPLETION_API, IMAGE_OUTPUT, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | IMAGE_GENERATION | (unknown) | (unknown) | (unknown)
GOOGLE | gemini-2.5-flash-lite | AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1048576 | 3600 per request GOOGLE | gemini-2.5-flash-lite | AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1048576 | 3600 per request | PROVIDER_API countTokens
GOOGLE | gemini-2.5-pro | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1048576 | 3600 per request GOOGLE | gemini-2.5-pro | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1048576 | 3600 per request | PROVIDER_API countTokens
GOOGLE | gemini-3-flash | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1000000 | 3600 per request GOOGLE | gemini-3-flash | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1000000 | 3600 per request | PROVIDER_API countTokens
GOOGLE | gemini-3-pro | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1000000 | 3600 per request GOOGLE | gemini-3-pro | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1000000 | 3600 per request | PROVIDER_API countTokens
GOOGLE | gemini-3-pro-image | ALWAYS_REASONING, CHAT_COMPLETION_API, IMAGE_OUTPUT, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | IMAGE_GENERATION | (unknown) | (unknown) GOOGLE | gemini-3-pro-image | ALWAYS_REASONING, CHAT_COMPLETION_API, IMAGE_OUTPUT, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | IMAGE_GENERATION | (unknown) | (unknown) | (unknown)
GOOGLE | gemini-3.1-flash-image | ALWAYS_REASONING, CHAT_COMPLETION_API, IMAGE_OUTPUT, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | IMAGE_GENERATION | (unknown) | (unknown) GOOGLE | gemini-3.1-flash-image | ALWAYS_REASONING, CHAT_COMPLETION_API, IMAGE_OUTPUT, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | IMAGE_GENERATION | (unknown) | (unknown) | (unknown)
GOOGLE | gemini-flash-latest | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1000000 | 3600 per request GOOGLE | gemini-flash-latest | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1000000 | 3600 per request | PROVIDER_API countTokens
GOOGLE | gemini-pro-latest | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1000000 | 3600 per request GOOGLE | gemini-pro-latest | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1000000 | 3600 per request | PROVIDER_API countTokens
GROQ | llama-3.3-70b-versatile | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) GROQ | llama-3.3-70b-versatile | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) | (unknown)
GROQ | moonshotai/kimi-k2-instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) GROQ | moonshotai/kimi-k2-instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
GROQ | openai/gpt-oss-120b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 131072 | (unknown) GROQ | openai/gpt-oss-120b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 131072 | (unknown) | (unknown)
GROQ | qwen/qwen3-32b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) GROQ | qwen/qwen3-32b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
GWDG | claude-sonnet-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request GWDG | claude-sonnet-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request | PROVIDER_API /v1/messages/count_tokens
GWDG | deepseek-r1 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) GWDG | deepseek-r1 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
GWDG | gemma-3-27b-it | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) GWDG | gemma-3-27b-it | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
GWDG | gpt-5.5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) GWDG | gpt-5.5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) | TIKTOKEN o200k_base
GWDG | internvl2.5-8b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) GWDG | internvl2.5-8b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
GWDG | meta-llama-3.1-8b-instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) GWDG | meta-llama-3.1-8b-instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) | (unknown)
GWDG | qwen3-235b-a22b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) GWDG | qwen3-235b-a22b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
HELMHOLTZ | 1 - Llama3 405 the best general model | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) HELMHOLTZ | 1 - Llama3 405 the best general model | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
HELMHOLTZ | 10 - Muse Glimmer 30b - the newest META model | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) HELMHOLTZ | 10 - Muse Glimmer 30b - the newest META model | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
HELMHOLTZ | Qwen 3.8-27B with DFlash on haicluster | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) HELMHOLTZ | Qwen 3.8-27B with DFlash on haicluster | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
HELMHOLTZ | alias-qwen38-27b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) HELMHOLTZ | alias-qwen38-27b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
HETZNER | gpt-oss-120b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 131072 | (unknown) HETZNER | gpt-oss-120b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 131072 | (unknown) | (unknown)
HETZNER | qwen3-coder-30b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) HETZNER | qwen3-coder-30b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
HUGGINGFACE | HuggingFaceTB/SmolLM3-3B | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) HUGGINGFACE | HuggingFaceTB/SmolLM3-3B | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
HUGGINGFACE | Qwen/Qwen3.8-27B | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) HUGGINGFACE | Qwen/Qwen3.8-27B | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
HUGGINGFACE | deepseek-ai/DeepSeek-R1-Distill-Qwen-32B | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) HUGGINGFACE | deepseek-ai/DeepSeek-R1-Distill-Qwen-32B | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
HUGGINGFACE | google/gemma-4-31B-it:novita | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) HUGGINGFACE | google/gemma-4-31B-it:novita | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
HUGGINGFACE | meta-llama/Llama-4-Scout-17B-16E-Instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) HUGGINGFACE | meta-llama/Llama-4-Scout-17B-16E-Instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
HUGGINGFACE | meta-llama/Meta-Llama-3.1-405B-Instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) HUGGINGFACE | meta-llama/Meta-Llama-3.1-405B-Instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) | (unknown)
HUGGINGFACE | mistralai/Magistral-Small-2509 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) HUGGINGFACE | mistralai/Magistral-Small-2509 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
HUGGINGFACE | openai/gpt-oss-120b:fireworks-ai | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 131072 | (unknown) HUGGINGFACE | openai/gpt-oss-120b:fireworks-ai | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 131072 | (unknown) | (unknown)
IONOS | meta-llama/Llama-3.3-70B-Instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) IONOS | meta-llama/Llama-3.3-70B-Instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) | (unknown)
IONOS | mistralai/Mistral-Small-24B-Instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) IONOS | mistralai/Mistral-Small-24B-Instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
LITE_LLM | anthropic/claude-sonnet-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request LITE_LLM | anthropic/claude-sonnet-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request | PROVIDER_API /v1/messages/count_tokens
LITE_LLM | azure/gpt-5.6 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) LITE_LLM | azure/gpt-5.6 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) | TIKTOKEN o200k_base
LITE_LLM | the-fast-one | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) LITE_LLM | the-fast-one | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
MISTRAL | codestral-2508 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) MISTRAL | codestral-2508 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
MISTRAL | magistral-medium-2506 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) MISTRAL | magistral-medium-2506 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
MISTRAL | ministral-14b-2512 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) MISTRAL | ministral-14b-2512 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
MISTRAL | ministral-3b-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) MISTRAL | ministral-3b-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
MISTRAL | ministral-8b-2410 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) MISTRAL | ministral-8b-2410 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
MISTRAL | mistral-large-2411 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) MISTRAL | mistral-large-2411 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) | (unknown)
MISTRAL | mistral-large-2512 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) MISTRAL | mistral-large-2512 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) | (unknown)
MISTRAL | mistral-large-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) MISTRAL | mistral-large-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) | (unknown)
MISTRAL | mistral-medium-2505 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) MISTRAL | mistral-medium-2505 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) | (unknown)
MISTRAL | mistral-medium-2508 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) MISTRAL | mistral-medium-2508 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) | (unknown)
MISTRAL | mistral-medium-2604 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) MISTRAL | mistral-medium-2604 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) | (unknown)
MISTRAL | mistral-medium-3-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) MISTRAL | mistral-medium-3-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) | (unknown)
MISTRAL | mistral-medium-3.5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) MISTRAL | mistral-medium-3.5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) | (unknown)
MISTRAL | mistral-medium-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) MISTRAL | mistral-medium-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) | (unknown)
MISTRAL | mistral-saba-2502 | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) MISTRAL | mistral-saba-2502 | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
MISTRAL | mistral-small-2501 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) MISTRAL | mistral-small-2501 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
MISTRAL | mistral-small-2503 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) MISTRAL | mistral-small-2503 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
MISTRAL | mistral-small-2603 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) MISTRAL | mistral-small-2603 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
MISTRAL | mistral-small-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) MISTRAL | mistral-small-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
MISTRAL | open-mistral-nemo | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) MISTRAL | open-mistral-nemo | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
MISTRAL | pixtral-12b-2409 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 128000 | (unknown) MISTRAL | pixtral-12b-2409 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 128000 | (unknown) | (unknown)
MISTRAL | pixtral-large-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 128000 | (unknown) MISTRAL | pixtral-large-latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 128000 | (unknown) | (unknown)
MISTRAL | voxtral-small-2507 | CHAT_COMPLETION_API, FUNCTION_CALLING, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT | TRANSCRIPTION | (unknown) | (unknown) MISTRAL | voxtral-small-2507 | CHAT_COMPLETION_API, FUNCTION_CALLING, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT | TRANSCRIPTION | (unknown) | (unknown) | (unknown)
NONE | gpt-5.6 | (nothing) | CHAT | (unknown) | (unknown) NONE | gpt-5.6 | (nothing) | CHAT | (unknown) | (unknown) | (unknown)
OPEN_AI | | (nothing) | CHAT | (unknown) | (unknown) OPEN_AI | | (nothing) | CHAT | (unknown) | (unknown) | (unknown)
OPEN_AI | gpt-3.5-turbo | RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) OPEN_AI | gpt-3.5-turbo | RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | TIKTOKEN cl100k_base
OPEN_AI | gpt-3.5-turbo-16k | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) OPEN_AI | gpt-3.5-turbo-16k | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | TIKTOKEN cl100k_base
OPEN_AI | gpt-4 | RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | 8192 | (unknown) OPEN_AI | gpt-4 | RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | 8192 | (unknown) | TIKTOKEN cl100k_base
OPEN_AI | gpt-4-0613 | RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | 8192 | (unknown) OPEN_AI | gpt-4-0613 | RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | 8192 | (unknown) | TIKTOKEN cl100k_base
OPEN_AI | gpt-4-turbo | FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | 128000 | (unknown) OPEN_AI | gpt-4-turbo | FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | 128000 | (unknown) | TIKTOKEN cl100k_base
OPEN_AI | gpt-4o | FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 128000 | (unknown) OPEN_AI | gpt-4o | FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 128000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-4o-audio-preview | FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | SPEECH_SYNTHESIS | 128000 | (unknown) OPEN_AI | gpt-4o-audio-preview | FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | SPEECH_SYNTHESIS | 128000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-4o-mini | FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 128000 | (unknown) OPEN_AI | gpt-4o-mini | FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 128000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-4o-mini-search-preview | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 128000 | (unknown) OPEN_AI | gpt-4o-mini-search-preview | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 128000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-4o-search-preview | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 128000 | (unknown) OPEN_AI | gpt-4o-search-preview | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 128000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-5 | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 400000 | (unknown) OPEN_AI | gpt-5 | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 400000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-5-mini | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 400000 | (unknown) OPEN_AI | gpt-5-mini | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 400000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-5-nano | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 400000 | (unknown) OPEN_AI | gpt-5-nano | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 400000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-5.1 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 400000 | (unknown) OPEN_AI | gpt-5.1 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 400000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-5.1-codex | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 400000 | (unknown) OPEN_AI | gpt-5.1-codex | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 400000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-5.2 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 400000 | (unknown) OPEN_AI | gpt-5.2 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 400000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-5.3 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) OPEN_AI | gpt-5.3 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-5.4 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) OPEN_AI | gpt-5.4 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-5.5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) OPEN_AI | gpt-5.5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-5.6 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) OPEN_AI | gpt-5.6 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | gpt-6-astra | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) OPEN_AI | gpt-6-astra | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) | (unknown)
OPEN_AI | gpt-6-astra-mini | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) OPEN_AI | gpt-6-astra-mini | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) | (unknown)
OPEN_AI | o1 | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | (unknown) OPEN_AI | o1 | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | o1-mini | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) OPEN_AI | o1-mini | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | TIKTOKEN o200k_base
OPEN_AI | o1-pro | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | (unknown) OPEN_AI | o1-pro | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | 200000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | o3 | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 200000 | (unknown) OPEN_AI | o3 | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 200000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | o3-mini | ALWAYS_REASONING, FUNCTION_CALLING, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) OPEN_AI | o3-mini | ALWAYS_REASONING, FUNCTION_CALLING, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | TIKTOKEN o200k_base
OPEN_AI | o3-pro | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 200000 | (unknown) OPEN_AI | o3-pro | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 200000 | (unknown) | TIKTOKEN o200k_base
OPEN_AI | o4-mini | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 200000 | (unknown) OPEN_AI | o4-mini | ALWAYS_REASONING, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, RESPONSES_API, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 200000 | (unknown) | TIKTOKEN o200k_base
OPEN_ROUTER | anthropic/claude-opus-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request OPEN_ROUTER | anthropic/claude-opus-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | 600 per request | PROVIDER_API /v1/messages/count_tokens
OPEN_ROUTER | deepseek/deepseek-chat-v3.1 | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) OPEN_ROUTER | deepseek/deepseek-chat-v3.1 | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
OPEN_ROUTER | deepseek/deepseek-r1-distill-llama-70b | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) OPEN_ROUTER | deepseek/deepseek-r1-distill-llama-70b | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
OPEN_ROUTER | google/gemini-3.7-flash | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1000000 | 3600 per request OPEN_ROUTER | google/gemini-3.7-flash | ALWAYS_REASONING, AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | 1000000 | 3600 per request | PROVIDER_API countTokens
OPEN_ROUTER | google/gemma-4-31b-it | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) OPEN_ROUTER | google/gemma-4-31b-it | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
OPEN_ROUTER | meta-llama/llama-4-maverick | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) OPEN_ROUTER | meta-llama/llama-4-maverick | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
OPEN_ROUTER | minimax/minimax-m2 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) OPEN_ROUTER | minimax/minimax-m2 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
OPEN_ROUTER | mistralai/mistral-large-3 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) OPEN_ROUTER | mistralai/mistral-large-3 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 256000 | (unknown) | (unknown)
OPEN_ROUTER | moonshotai/kimi-k2-thinking | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) OPEN_ROUTER | moonshotai/kimi-k2-thinking | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
OPEN_ROUTER | nvidia/nemotron-3-49b | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) OPEN_ROUTER | nvidia/nemotron-3-49b | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
OPEN_ROUTER | openai/gpt-5.6 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) OPEN_ROUTER | openai/gpt-5.6 | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 1050000 | (unknown) | TIKTOKEN o200k_base
OPEN_ROUTER | openai/gpt-oss-120b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 131072 | (unknown) OPEN_ROUTER | openai/gpt-oss-120b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 131072 | (unknown) | (unknown)
OPEN_ROUTER | perplexity/sonar-reasoning | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) OPEN_ROUTER | perplexity/sonar-reasoning | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) | (unknown)
OPEN_ROUTER | qwen/qwen3.8-flash-next | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) OPEN_ROUTER | qwen/qwen3.8-flash-next | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) | (unknown)
OPEN_ROUTER | z-ai/glm-5.3 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) OPEN_ROUTER | z-ai/glm-5.3 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
PERPLEXITY | sonar | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) PERPLEXITY | sonar | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) | (unknown)
PERPLEXITY | sonar-deep-research | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) PERPLEXITY | sonar-deep-research | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) | (unknown)
PERPLEXITY | sonar-pro | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) PERPLEXITY | sonar-pro | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) | (unknown)
PERPLEXITY | sonar-reasoning | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) PERPLEXITY | sonar-reasoning | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) | (unknown)
PERPLEXITY | sonar-reasoning-pro | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) PERPLEXITY | sonar-reasoning-pro | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | | (nothing) | CHAT | (unknown) | (unknown) SELF_HOSTED | | (nothing) | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | --- | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | --- | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | 01-ai/yi-large | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | 01-ai/yi-large | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | a-model-nobody-has-heard-of | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | a-model-nobody-has-heard-of | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | apertus-1.5-8b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | apertus-1.5-8b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | apriel-1.5-15b-thinker | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | apriel-1.5-15b-thinker | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | apriel-1.6-15b-thinker | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | apriel-1.6-15b-thinker | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | aya-expanse:8b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | aya-expanse:8b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | aya-vision:8b | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | aya-vision:8b | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | command-a-plus | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | command-a-plus | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | command-a-reasoning | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | command-a-reasoning | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | command-a-vision | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | command-a-vision | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | command-a:111b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | command-a:111b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | command-r7b:7b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | command-r7b:7b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | deepseek-r1-distill-llama-70b | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | deepseek-r1-distill-llama-70b | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | deepseek-r1:32b | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | deepseek-r1:32b | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | deepseek-v2.5 | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | deepseek-v2.5 | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | deepseek-v3.1:671b | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | deepseek-v3.1:671b | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | ernie-4.5-21b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | ernie-4.5-21b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | ernie-4.5-vl-28b | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | ernie-4.5-vl-28b | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | ernie-x1.1-thinking | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | ernie-x1.1-thinking | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | eurollm-9b-instruct | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | eurollm-9b-instruct | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | falcon-h1-1.5b-tool-calling | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | falcon-h1-1.5b-tool-calling | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | falcon-h1:7b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | falcon-h1:7b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | falcon3:10b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | falcon3:10b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | gemma2:9b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | gemma2:9b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | gemma3:1b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | gemma3:1b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | gemma3:27b | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | gemma3:27b | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | gemma3n:e4b | AUDIO_INPUT, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | gemma3n:e4b | AUDIO_INPUT, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | gemma4:31b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | gemma4:31b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | gemma4:e2b | AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | gemma4:e2b | AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | glm-4-9b-chat | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | glm-4-9b-chat | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | glm-4.5v | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | glm-4.5v | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | glm-4.6:latest | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | glm-4.6:latest | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | glm-5-2 | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | glm-5-2 | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | glm-5.3-flash-nvfp4 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | glm-5.3-flash-nvfp4 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | gpt-oss:20b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 131072 | (unknown) SELF_HOSTED | gpt-oss:20b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT, WEB_SEARCH | CHAT | 131072 | (unknown) | (unknown)
SELF_HOSTED | granite3.2-vision:2b | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | granite3.2-vision:2b | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | hunyuan:7b | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | hunyuan:7b | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | inclusionai/ling-mini-2.0 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | inclusionai/ling-mini-2.0 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | internlm3:8b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | internlm3:8b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | internvl3-8b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | internvl3-8b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | kimi-k2.7-code | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | kimi-k2.7-code | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | kimi-k2:1t | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | kimi-k2:1t | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | kimi-k3:latest | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | kimi-k3:latest | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT, VIDEO_INPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | kimi-vl:16b | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | kimi-vl:16b | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | ling-1t | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | ling-1t | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | llama-3.1-405b-base | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) SELF_HOSTED | llama-3.1-405b-base | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) | (unknown)
SELF_HOSTED | llama2:13b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | llama2:13b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | llama3.2-vision:11b | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | llama3.2-vision:11b | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | llama3.2:3b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) SELF_HOSTED | llama3.2:3b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | 131072 | (unknown) | (unknown)
SELF_HOSTED | magistral:24b | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | magistral:24b | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | minimax-m2:latest | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | minimax-m2:latest | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | minimax-text-01 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | minimax-text-01 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | ministral-8b-instruct-2410 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | ministral-8b-instruct-2410 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | mistral-nemo:12b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | mistral-nemo:12b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | mistral-small3.2:24b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | mistral-small3.2:24b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | muse-glimmer-30b | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | muse-glimmer-30b | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | nemotron-3-49b | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | nemotron-3-49b | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | nvidia-nemotron-3.5-lightning-30b-a3b-nvfp4 | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | nvidia-nemotron-3.5-lightning-30b-a3b-nvfp4 | CHAT_COMPLETION_API, FUNCTION_CALLING, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | occiglot-7b-eu5 | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | occiglot-7b-eu5 | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | olmo-3-32b-think | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | olmo-3-32b-think | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | olmo2:13b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | olmo2:13b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | olmo3:7b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | olmo3:7b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | phi-4-mini-reasoning | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | phi-4-mini-reasoning | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | phi-4-multimodal-instruct | AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | phi-4-multimodal-instruct | AUDIO_INPUT, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | phi-4-reasoning-vision | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | phi-4-reasoning-vision | ALWAYS_REASONING, CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | phi3:14b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | phi3:14b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | phi4-mini:latest | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | phi4-mini:latest | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | phi4:14b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | phi4:14b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | qwen2.5-vl-7b-instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | qwen2.5-vl-7b-instruct | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | qwen3-coder:30b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | qwen3-coder:30b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | qwen3.5:32b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | qwen3.5:32b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | qwen3.6:32b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | qwen3.6:32b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | qwen3.8-2.4t-a95b | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | qwen3.8-2.4t-a95b | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | qwen3.8:27b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | qwen3.8:27b | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | qwen3.8:27b-mlx | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | qwen3.8:27b-mlx | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | qwen3.8:latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | qwen3.8:latest | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, REASONING_BY_DEFAULT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | qwq:32b | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | qwq:32b | ALWAYS_REASONING, CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | ring-1t | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | ring-1t | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | salamandra-7b-instruct | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | salamandra-7b-instruct | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | salamandra-7b-instruct-tools | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | salamandra-7b-instruct-tools | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | seed-oss:36b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | seed-oss:36b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | smollm2:1.7b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | smollm2:1.7b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | smollm3:3b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | smollm3:3b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | starling-lm:7b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | starling-lm:7b | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | tencent/hy3 | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | tencent/hy3 | CHAT_COMPLETION_API, FUNCTION_CALLING, OPTIONAL_REASONING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | teuken-7b-instruct | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | teuken-7b-instruct | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
SELF_HOSTED | voxtral-mini-3b | CHAT_COMPLETION_API, FUNCTION_CALLING, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT | TRANSCRIPTION | (unknown) | (unknown) SELF_HOSTED | voxtral-mini-3b | CHAT_COMPLETION_API, FUNCTION_CALLING, SPEECH_INPUT, TEXT_INPUT, TEXT_OUTPUT | TRANSCRIPTION | (unknown) | (unknown) | (unknown)
SELF_HOSTED | yi-1.5:9b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) SELF_HOSTED | yi-1.5:9b | CHAT_COMPLETION_API, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
X | grok-2-vision-1212 | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) X | grok-2-vision-1212 | CHAT_COMPLETION_API, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
X | grok-3 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) X | grok-3 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
X | grok-3-mini | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) X | grok-3-mini | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
X | grok-4 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) X | grok-4 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
X | grok-4-fast-reasoning | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) X | grok-4-fast-reasoning | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)
X | grok-4.20 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | (unknown) X | grok-4.20 | ALWAYS_REASONING, CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | (unknown) | (unknown)
X | grok-4.20-non-reasoning | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | (unknown) X | grok-4.20-non-reasoning | CHAT_COMPLETION_API, FUNCTION_CALLING, MULTIPLE_IMAGE_INPUT, TEXT_INPUT, TEXT_OUTPUT | CHAT | 1000000 | (unknown) | (unknown)
X | grok-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) X | grok-5 | CHAT_COMPLETION_API, FUNCTION_CALLING, TEXT_INPUT, TEXT_OUTPUT | CHAT | (unknown) | (unknown) | (unknown)

106
app/Tests/Models/TokenizerRuleTests.cs Normal file
View File

@ -0,0 +1,106 @@
using AIStudio.Models;
using AIStudio.Models.Registry;
using AIStudio.Provider;
using AIStudio.Settings;
namespace AIStudio.Tests.Models;
/// <summary>
/// Checks which tokenizer the rules name for a model.
/// </summary>
/// <remarks>
/// Naming one changes nothing about the counting today: AI Studio counts with the tokenizer it
/// ships unless somebody points it at a tokenizer.json file, and none of the references below is
/// such a file. What they are for is the sentence in the provider dialog, which until now let a
/// person guess -- including the ones who go looking for a file which was never published.
///
/// The kind matters as much as the name, and that is what the cases here pin. "o200k_base" is an
/// encoding nobody can download, "/v1/messages/count_tokens" is an endpoint nobody can select in a
/// file dialog, and telling them apart is the whole point of recording the kind alongside the name.
/// </remarks>
[TestFixture]
public sealed class TokenizerRuleTests
{
[TestCase(LLMProviders.OPEN_AI, "gpt-5.1", "o200k_base")]
[TestCase(LLMProviders.OPEN_AI, "gpt-5.6", "o200k_base", Description = "Every model of the 5 line inherits the encoding of its prefix.")]
[TestCase(LLMProviders.OPEN_AI, "gpt-5-chat-latest", "o200k_base")]
[TestCase(LLMProviders.OPEN_AI, "gpt-4o", "o200k_base")]
[TestCase(LLMProviders.OPEN_AI, "gpt-4o-mini-search-preview", "o200k_base", Description = "Stated in full rather than inherited, so it has to say this itself.")]
[TestCase(LLMProviders.OPEN_AI, "o1", "o200k_base")]
[TestCase(LLMProviders.OPEN_AI, "o1-mini", "o200k_base")]
[TestCase(LLMProviders.OPEN_AI, "o3-mini", "o200k_base")]
[TestCase(LLMProviders.OPEN_AI, "o4-mini", "o200k_base")]
[TestCase(LLMProviders.OPEN_AI, "gpt-4", "cl100k_base", Description = "The older encoding, which is where the 4 line stayed.")]
[TestCase(LLMProviders.OPEN_AI, "gpt-4-turbo", "cl100k_base")]
[TestCase(LLMProviders.OPEN_AI, "gpt-3.5-turbo", "cl100k_base")]
public void OpenAINamesAnEncodingRatherThanAFile(LLMProviders provider, string modelId, string encoding)
{
var tokenizer = provider.GetModelProfile(new Model(modelId, null)).Tokenizer;
Assert.Multiple(() =>
{
Assert.That(tokenizer.Kind, Is.EqualTo(TokenizerKind.TIKTOKEN));
Assert.That(tokenizer.Id, Is.EqualTo(encoding));
});
}
[TestCase(LLMProviders.ANTHROPIC, "claude-opus-5", "/v1/messages/count_tokens")]
[TestCase(LLMProviders.ANTHROPIC, "claude-3-5-haiku-latest", "/v1/messages/count_tokens")]
[TestCase(LLMProviders.GOOGLE, "gemini-3-pro", "countTokens")]
[TestCase(LLMProviders.GOOGLE, "gemini-2.5-flash-lite", "countTokens")]
public void AnthropicAndGoogleNameAnEndpointBecauseTheyPublishNoFile(LLMProviders provider, string modelId, string endpoint)
{
var tokenizer = provider.GetModelProfile(new Model(modelId, null)).Tokenizer;
Assert.Multiple(() =>
{
Assert.That(tokenizer.Kind, Is.EqualTo(TokenizerKind.PROVIDER_API));
Assert.That(tokenizer.Id, Is.EqualTo(endpoint));
});
}
[TestCase(LLMProviders.OPEN_AI, "gpt-6-astra", Description = "Newer than the mapping OpenAI publishes, so nothing is claimed for it.")]
[TestCase(LLMProviders.OPEN_AI, "gpt-oss-120b", Description = "Open weights, and not in OpenAI's encoding table either.")]
[TestCase(LLMProviders.SELF_HOSTED, "some-model-nobody-wrote-a-rule-for")]
[TestCase(LLMProviders.MISTRAL, "mistral-large-2512")]
public void AModelNobodyNamedATokenizerForSaysSo(LLMProviders provider, string modelId)
{
var tokenizer = provider.GetModelProfile(new Model(modelId, null)).Tokenizer;
Assert.Multiple(() =>
{
Assert.That(tokenizer.IsKnown, Is.False);
Assert.That(tokenizer.Kind, Is.EqualTo(TokenizerKind.UNKNOWN), "Which means the built-in tokenizer, the same as today.");
});
}
[Test]
public void TheSameModelThroughAGatewayKeepsItsTokenizer()
{
//
// A gateway cuts what its transport cannot carry, which is about APIs. Which tokenizer a
// model was trained with is a property of the model and survives the trip.
//
var directly = ModelRegistry.Shared.Profile(LLMProviders.OPEN_AI, "gpt-5.1");
var throughAGateway = ModelRegistry.Shared.Profile(LLMProviders.OPEN_ROUTER, "openai/gpt-5.1");
Assert.That(throughAGateway.Tokenizer, Is.EqualTo(directly.Tokenizer));
}
[Test]
public void ANameWithoutAKindIsNotAReference()
{
//
// Both halves have to be there. A kind without a name says nothing to act on, and a name
// without a kind cannot be told apart from any other string -- whether it is a repository,
// an encoding or an endpoint decides what a person can do with it.
//
Assert.Multiple(() =>
{
Assert.That(new TokenizerRef(TokenizerKind.HUGGING_FACE, string.Empty).IsKnown, Is.False);
Assert.That(new TokenizerRef(TokenizerKind.UNKNOWN, "o200k_base").IsKnown, Is.False);
Assert.That(TokenizerRef.UNKNOWN.IsKnown, Is.False);
Assert.That(new TokenizerRef(TokenizerKind.TIKTOKEN, "o200k_base").IsKnown, Is.True);
});
}
}