mirror of
https://github.com/MindWorkAI/AI-Studio.git
synced 2026-08-17 14:52:11 +00:00
e0d7a810ea
Add LiteLLM as a first-class provider. LiteLLM is a self-hosted AI gateway that exposes a single OpenAI-compatible endpoint in front of 100+ models across providers (OpenAI, Anthropic, Google, Azure, AWS Bedrock, and more). The provider mirrors the existing OpenAI-compatible providers: it uses the shared StreamOpenAICompatibleChatCompletion helper for streaming chat and LoadModelsResponse against /v1/models for automatic model discovery. The user supplies the base URL of their proxy; it is normalized and the OpenAI-compatible /v1/ path is appended. It uses the self-hosted trust policy and confidence handling, since the gateway owner decides which downstream providers requests are routed to. Wired through the standard plumbing: the LLMProviders enum, LLMProvidersExtensions (name, confidence, factory, hostname/API-key requirements), ProviderExtensions (capabilities + reasoning), and SettingsManager (per-scheme confidence levels). Hostname and API key are both required. Added a user-facing changelog entry and updated the configuration-plugin docs to include the new LITE_LLM confidence key. |
||
|---|---|---|
| .. | ||
| changelog | ||
| fonts | ||
| sounds | ||
| specs/eri | ||
| svg | ||
| system | ||
| app.css | ||
| app.js | ||
| audio-recorder-worklet.js | ||
| audio.js | ||
| boot.js | ||
| chat-math.js | ||
| diff.js | ||