Tell embedding providers which tokenizer their model uses

app/MindWork AI Studio/Models/OpenAI/OpenAIEmbeddingFamily.cs

@@ -23,12 +23,19 @@ public sealed class OpenAIEmbeddingFamily : ModelFamily
     /// <inheritdoc />
     public override ModelSource Source => new("https://platform.openai.com/docs/guides/embeddings", new DateOnly(2026, 9, 11), "The app lists these under IProvider.GetEmbeddingModels, which is where the statement that they embed comes from.");
 
+    /// <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 names all three of these models -- text-embedding-3-small, text-embedding-3-large and text-embedding-ada-002 -- and maps every one of them to cl100k_base rather than to the newer o200k_base of the chat models.")
+    ];
+
     /// <inheritdoc />
     protected override void Declare(ModelFamilyBuilder builder)
     {
         builder.Rule("text-embedding-3").AsPrefix()
             .Capabilities(TEXT_INPUT | EMBEDDING)
-            .Kind(ModelKind.EMBEDDING);
+            .Kind(ModelKind.EMBEDDING)
+            .Tokenizer(TokenizerKind.TIKTOKEN, "cl100k_base");
 
         builder.Rule("text-embedding-ada").AsPrefix().Inherits();
     }