[Bug]: OpenAI provider prefix stripped twice for namespaced model IDs in Responses API

Open Beginner friendly
#4,613 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Assessment

Difficulty
2/5
Estimated time
1-3 hours
Newbie friendliness
78/100
Issue type
Bug
Clarity
Clearly specified
Activity status
Active
Tech stack
python
Domain
backend-api-design, testing-qa

Research direction

Start with LLMProvider.as_litellm_call_kwargs() and run tests/sdk/llm/test_litellm_provider.py::test_llm_provider_preserves_namespaced_model_after_litellm_reparse. Review the existing nested-provider tests, then verify the full test file and confirm that namespaced models retain the correct identifier after LiteLLM reparses them.

Written by the indexing model from the issue text.

Description

bug llm priority:medium ready-for-dev sdk
Bug Description

LLMProvider.from_model() parses a model string like openai/openai/example-model into model="openai/example-model" + provider="openai". When as_litellm_call_kwargs() forwards that as model="openai/example-model" + custom_llm_provider="openai", LiteLLM's Responses API path re-parses and strips the openai/ prefix again, sending example-model upstream and causing 400 Model not found.

This also affects any namespaced model ID whose inner segment is a LiteLLM provider name (e.g. openai/openai/o3-mini, mistral/mistral/large).

Upstream report: OpenHands/OpenHands#16365

Expected Behavior

The SDK should pass a model string that, after LiteLLM re-parses it, resolves to the originally parsed openai/example-model.

Actual Behavior

Before the fix, running the repro path through as_litellm_call_kwargs() sends example-model to the upstream endpoint. The upstream returns 400 Model not found because the model identifier has lost its provider namespace.

uv run pytest tests/sdk/llm/test_litellm_provider.py::test_llm_provider_preserves_namespaced_model_after_litellm_reparse -v

The regression test above (introduced in the fix) fails on main and passes after the change.

Acceptance Criteria
  • LLMProvider.as_litellm_call_kwargs() re-adds the provider prefix when self.model already starts with it, so LiteLLM strips exactly one prefix and lands on the correct parsed model.
  • uv run pytest tests/sdk/llm/test_litellm_provider.py -v passes.
  • Existing nested-provider cases (openrouter, bedrock, mistral, litellm_proxy) continue to pass.
Dominant language
Python
Stars
1.1k
Forks
542
Avg merge
1d 22h
Merged PRs (30d)
139

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from OpenHands/software-agent-sdk

All issues in OpenHands/software-agent-sdk

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.