mirror of
https://github.com/BerriAI/litellm.git
synced 2026-11-17 09:34:48 +00:00
|
Some checks failed
Cost map pins recheck / rerun-cost-map-guard (push) Has been cancelled
LiteLLM Rust DB / db-queries (push) Has been cancelled
LiteLLM Rust / rust-lint (push) Has been cancelled
LiteLLM Rust / rust-test (push) Has been cancelled
LiteLLM Rust / rust-wheel (push) Has been cancelled
Terraform Provider / gofmt, vet, build, test (push) Has been cancelled
Terraform Provider / Provider endpoints vs proxy OpenAPI schema (push) Has been cancelled
Each client surface (/v1/chat/completions, /v1/responses, /v1/messages) now decides the provider's native API format once with litellm/native/route_plan.py, owns logging for the request, and hands the translated request to a native executor (execute_chat, execute_responses, execute_messages) instead of calling back into a public entry point. - Messages to Responses, Messages to chat, Responses to chat and chat to Responses bridges call the native executors - Responses-mode models on providers with no Responses API stay on chat - Bridges price the provider reply before translating it, keep its headers, map provider errors and run JSON-schema validation as re-entry did - Bridged streams are logged once, under the client's call type, when the client stream ends - execute_chat and the chat provider helpers move from main.py to litellm/native/, and a lint check keeps litellm.native from importing litellm.main |
||
|---|---|---|
| .. | ||
| embedding | ||
| messages | ||
| responses | ||
| __init__.py | ||
| test_github_copilot_authenticator.py | ||
| test_github_copilot_transformation.py | ||
| test_github_host.py | ||
| test_per_user_auth.py | ||