ouyangbo/EvoScientist-Multi
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(llm): upgrade OpenAI reasoning effort from high to xhigh (#136)

Browse Source 
* feat(llm): upgrade OpenAI reasoning effort from high to xhigh

The OpenAI Responses API supports "xhigh" as a reasoning effort level,
which provides deeper reasoning than "high". This is already used by
other CLI tools (e.g., OpenClaw) for OpenAI models.

Only affects the direct API key path; the ccproxy/OAuth path is
unchanged (reasoning is still skipped there).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix(llm): limit xhigh reasoning to gpt-5.4+ and codex models

Only gpt-5.4 series and codex models support xhigh reasoning effort.
Older models (gpt-5, gpt-5.1, gpt-5.2, gpt-5.3) fall back to high.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: Xi Zhang <zacharyzhang2022@gmail.com>
Co-authored-by: Xi Zhang <106144707+X-iZhang@users.noreply.github.com>
This commit is contained in:
Yinhan Lu
2026-04-03 09:35:14 -04:00
committed by GitHub
parent 91de173c78
commit 5aa8353613
2 changed files with 31 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
EvoScientist/llm/models.py
+2 -1
View File
@@ -215,7 +215,8 @@ def _apply_auto_config(
# ccproxy uses Chat Completions which doesn't support reasoning.
pass
else:
kwargs["reasoning"] = {"effort": "high", "summary": "auto"}
_eff = "xhigh" if ("5.4" in model_id or "codex" in model_id) else "high"
kwargs["reasoning"] = {"effort": _eff, "summary": "auto"}
# Google GenAI: surface thinking traces
if provider == "google-genai":
tests/test_llm.py
+29 -4
View File
@@ -771,15 +771,40 @@ class TestAutoConfig:
assert call_kwargs["thinking"] == custom_thinking
@patch("EvoScientist.llm.models.init_chat_model")
def test_openai_reasoning(self, mock_init, monkeypatch):
"""Native OpenAI models get auto-reasoning."""
def test_openai_reasoning_xhigh(self, mock_init, monkeypatch):
"""gpt-5.4+ and codex models get xhigh reasoning."""
mock_init.return_value = "mock_model"
monkeypatch.delenv("OPENAI_BASE_URL", raising=False)
get_chat_model("gpt-5.4", provider="openai")
assert mock_init.call_args[1]["reasoning"] == {
"effort": "xhigh",
"summary": "auto",
}
get_chat_model("gpt-5.3-codex", provider="openai")
assert mock_init.call_args[1]["reasoning"] == {
"effort": "xhigh",
"summary": "auto",
}
@patch("EvoScientist.llm.models.init_chat_model")
def test_openai_reasoning_high_fallback(self, mock_init, monkeypatch):
"""Other OpenAI models get high reasoning effort."""
mock_init.return_value = "mock_model"
monkeypatch.delenv("OPENAI_BASE_URL", raising=False)
get_chat_model("gpt-5-nano")
assert mock_init.call_args[1]["reasoning"] == {
"effort": "high",
"summary": "auto",
}
call_kwargs = mock_init.call_args[1]
assert call_kwargs["reasoning"] == {"effort": "high", "summary": "auto"}
get_chat_model("gpt-5.2", provider="openai")
assert mock_init.call_args[1]["reasoning"] == {
"effort": "high",
"summary": "auto",
}
@patch("EvoScientist.llm.models.init_chat_model")
def test_openai_base_url_override(self, mock_init, monkeypatch):