swiftllm v0.2.0

New Providers

Google Gemini — full support for Gemini 2.0 Flash, 2.0 Pro, 1.5 Pro, and 1.5 Flash. Translates to Gemini's generateContent API with system instruction handling and SSE streaming.
Mistral AI — support for Mistral Large, Mistral Small, Codestral, Ministral 8B, and Pixtral Large. Uses Mistral's OpenAI-compatible API with full streaming.

Updated Model Support

OpenAI — added routing and cost tracking for gpt-4.1, gpt-4.1-mini, gpt-4.1-nano, o3, o4-mini, and chatgpt-* models.

How It Works
All new providers use the same unified API — just set the model name and swiftllm routes it automatically:
bashcurl http://localhost:8080/v1/chat/completions
-H "Authorization: Bearer your-key"
-d '{"model": "gemini-2.0-flash", "messages": [{"role": "user", "content": "Hello!"}]}'
Prefix routing: gemini-* → Google, mistral-* / codestral* / pixtral* → Mistral.

Full Changelog: v0.1.0...v0.2.0