mirror of
https://github.com/BerriAI/litellm.git
synced 2026-10-26 06:03:19 +00:00
b31b55075b
The Responses route now resolves xai and hosted_vllm. The Python host fills in the key and base get_llm_provider returns before the provider config reads its own defaults, as Python's responses entry point does. xAI drops metadata, reshapes code_interpreter, web_search and x_search tools, keeps max_output_tokens as sent and restates usage.cost from cost_in_usd_ticks. Hosted vLLM sends to /v1/responses with the fake-api-key fallback. xAI OAuth and hosted vLLM without an api_base decline to Python |
||
|---|---|---|
| .. | ||
| cli | ||
| shared | ||
| strategies | ||
| __init__.py | ||
| __main__.py | ||
| AGENTS.md | ||
| conftest.py | ||