mirror of
https://github.com/BerriAI/litellm.git
synced 2026-09-22 00:31:44 +00:00
fix: add gpt-5.1-mini and gpt-5.1-nano registry entries with supports_none_reasoning_effort
Per OpenAI's parameter compatibility guide, the gpt-5.1 family
(including gpt-5.1-mini and gpt-5.1-nano) supports `temperature` when
reasoning_effort is omitted or set to 'none'. LiteLLM's pre-flight
check correctly applies this for the bare `gpt-5.1` slug (which has
`supports_none_reasoning_effort: true` in the registry), but
`gpt-5.1-mini` and `gpt-5.1-nano` are absent from
`model_prices_and_context_window.json` entirely, so they fall back to
strict rejection — `litellm.UnsupportedParamsError` even when
reasoning_effort is null.
Adds the two missing entries with the same flag matrix as the existing
`gpt-5.1` entry — most importantly `supports_none_reasoning_effort:
true`. Pricing is anchored to the gpt-5-mini and gpt-5-nano siblings
(same scale tier, same provider) until OpenAI publishes explicit 5.1
mini/nano pricing; this matches the conservative "match the 5.0
generation siblings" rule rather than inventing numbers. Context
window (272000 input / 128000 output) matches the rest of the gpt-5.1
family.
Also extends two existing tests in tests/test_litellm/llms/openai/
test_gpt5_transformation.py:
- test_gpt5_1_model_detection: asserts gpt-5.1-mini and
gpt-5.1-nano now resolve through
`_supports_reasoning_effort_level(model, "none")` as True.
- test_gpt5_1_mini_nano_temperature_with_reasoning_effort_none:
new test mirroring the existing
test_gpt5_1_temperature_with_reasoning_effort_none, asserting
temperature pass-through for these two new slugs.
Repro and full curl evidence in #27351.
This commit is contained in:
Daniel Naves de Carvalho
parent
50ef2d51a2
commit
9268dfc08a
3 changed files with 187 additions and 0 deletions
|
|
@ -20746,6 +20746,89 @@
|
|||
"supports_xhigh_reasoning_effort": false,
|
||||
"supports_minimal_reasoning_effort": true
|
||||
},
|
||||
"gpt-5.1-mini": {
|
||||
"cache_read_input_token_cost": 2.5e-08,
|
||||
"cache_read_input_token_cost_flex": 1.25e-08,
|
||||
"cache_read_input_token_cost_priority": 4.5e-08,
|
||||
"input_cost_per_token": 2.5e-07,
|
||||
"input_cost_per_token_flex": 1.25e-07,
|
||||
"input_cost_per_token_priority": 4.5e-07,
|
||||
"litellm_provider": "openai",
|
||||
"max_input_tokens": 272000,
|
||||
"max_output_tokens": 128000,
|
||||
"max_tokens": 128000,
|
||||
"mode": "chat",
|
||||
"output_cost_per_token": 2e-06,
|
||||
"output_cost_per_token_flex": 1e-06,
|
||||
"output_cost_per_token_priority": 3.6e-06,
|
||||
"supported_endpoints": [
|
||||
"/v1/chat/completions",
|
||||
"/v1/batch",
|
||||
"/v1/responses"
|
||||
],
|
||||
"supported_modalities": [
|
||||
"text",
|
||||
"image"
|
||||
],
|
||||
"supported_output_modalities": [
|
||||
"text"
|
||||
],
|
||||
"supports_function_calling": true,
|
||||
"supports_native_streaming": true,
|
||||
"supports_parallel_function_calling": true,
|
||||
"supports_pdf_input": true,
|
||||
"supports_prompt_caching": true,
|
||||
"supports_reasoning": true,
|
||||
"supports_response_schema": true,
|
||||
"supports_system_messages": true,
|
||||
"supports_tool_choice": true,
|
||||
"supports_service_tier": true,
|
||||
"supports_vision": true,
|
||||
"supports_web_search": true,
|
||||
"supports_none_reasoning_effort": true,
|
||||
"supports_xhigh_reasoning_effort": false,
|
||||
"supports_minimal_reasoning_effort": true
|
||||
},
|
||||
"gpt-5.1-nano": {
|
||||
"cache_read_input_token_cost": 5e-09,
|
||||
"cache_read_input_token_cost_flex": 2.5e-09,
|
||||
"input_cost_per_token": 5e-08,
|
||||
"input_cost_per_token_flex": 2.5e-08,
|
||||
"input_cost_per_token_priority": 2.5e-06,
|
||||
"litellm_provider": "openai",
|
||||
"max_input_tokens": 272000,
|
||||
"max_output_tokens": 128000,
|
||||
"max_tokens": 128000,
|
||||
"mode": "chat",
|
||||
"output_cost_per_token": 4e-07,
|
||||
"output_cost_per_token_flex": 2e-07,
|
||||
"supported_endpoints": [
|
||||
"/v1/chat/completions",
|
||||
"/v1/batch",
|
||||
"/v1/responses"
|
||||
],
|
||||
"supported_modalities": [
|
||||
"text",
|
||||
"image"
|
||||
],
|
||||
"supported_output_modalities": [
|
||||
"text"
|
||||
],
|
||||
"supports_function_calling": true,
|
||||
"supports_native_streaming": true,
|
||||
"supports_parallel_function_calling": true,
|
||||
"supports_pdf_input": true,
|
||||
"supports_prompt_caching": true,
|
||||
"supports_reasoning": true,
|
||||
"supports_response_schema": true,
|
||||
"supports_system_messages": true,
|
||||
"supports_tool_choice": true,
|
||||
"supports_vision": true,
|
||||
"supports_web_search": true,
|
||||
"supports_none_reasoning_effort": true,
|
||||
"supports_xhigh_reasoning_effort": false,
|
||||
"supports_minimal_reasoning_effort": true
|
||||
},
|
||||
"gpt-5.2-codex": {
|
||||
"cache_read_input_token_cost": 1.75e-07,
|
||||
"cache_read_input_token_cost_priority": 3.5e-07,
|
||||
|
|
|
|||
|
|
@ -20760,6 +20760,89 @@
|
|||
"supports_xhigh_reasoning_effort": false,
|
||||
"supports_minimal_reasoning_effort": true
|
||||
},
|
||||
"gpt-5.1-mini": {
|
||||
"cache_read_input_token_cost": 2.5e-08,
|
||||
"cache_read_input_token_cost_flex": 1.25e-08,
|
||||
"cache_read_input_token_cost_priority": 4.5e-08,
|
||||
"input_cost_per_token": 2.5e-07,
|
||||
"input_cost_per_token_flex": 1.25e-07,
|
||||
"input_cost_per_token_priority": 4.5e-07,
|
||||
"litellm_provider": "openai",
|
||||
"max_input_tokens": 272000,
|
||||
"max_output_tokens": 128000,
|
||||
"max_tokens": 128000,
|
||||
"mode": "chat",
|
||||
"output_cost_per_token": 2e-06,
|
||||
"output_cost_per_token_flex": 1e-06,
|
||||
"output_cost_per_token_priority": 3.6e-06,
|
||||
"supported_endpoints": [
|
||||
"/v1/chat/completions",
|
||||
"/v1/batch",
|
||||
"/v1/responses"
|
||||
],
|
||||
"supported_modalities": [
|
||||
"text",
|
||||
"image"
|
||||
],
|
||||
"supported_output_modalities": [
|
||||
"text"
|
||||
],
|
||||
"supports_function_calling": true,
|
||||
"supports_native_streaming": true,
|
||||
"supports_parallel_function_calling": true,
|
||||
"supports_pdf_input": true,
|
||||
"supports_prompt_caching": true,
|
||||
"supports_reasoning": true,
|
||||
"supports_response_schema": true,
|
||||
"supports_system_messages": true,
|
||||
"supports_tool_choice": true,
|
||||
"supports_service_tier": true,
|
||||
"supports_vision": true,
|
||||
"supports_web_search": true,
|
||||
"supports_none_reasoning_effort": true,
|
||||
"supports_xhigh_reasoning_effort": false,
|
||||
"supports_minimal_reasoning_effort": true
|
||||
},
|
||||
"gpt-5.1-nano": {
|
||||
"cache_read_input_token_cost": 5e-09,
|
||||
"cache_read_input_token_cost_flex": 2.5e-09,
|
||||
"input_cost_per_token": 5e-08,
|
||||
"input_cost_per_token_flex": 2.5e-08,
|
||||
"input_cost_per_token_priority": 2.5e-06,
|
||||
"litellm_provider": "openai",
|
||||
"max_input_tokens": 272000,
|
||||
"max_output_tokens": 128000,
|
||||
"max_tokens": 128000,
|
||||
"mode": "chat",
|
||||
"output_cost_per_token": 4e-07,
|
||||
"output_cost_per_token_flex": 2e-07,
|
||||
"supported_endpoints": [
|
||||
"/v1/chat/completions",
|
||||
"/v1/batch",
|
||||
"/v1/responses"
|
||||
],
|
||||
"supported_modalities": [
|
||||
"text",
|
||||
"image"
|
||||
],
|
||||
"supported_output_modalities": [
|
||||
"text"
|
||||
],
|
||||
"supports_function_calling": true,
|
||||
"supports_native_streaming": true,
|
||||
"supports_parallel_function_calling": true,
|
||||
"supports_pdf_input": true,
|
||||
"supports_prompt_caching": true,
|
||||
"supports_reasoning": true,
|
||||
"supports_response_schema": true,
|
||||
"supports_system_messages": true,
|
||||
"supports_tool_choice": true,
|
||||
"supports_vision": true,
|
||||
"supports_web_search": true,
|
||||
"supports_none_reasoning_effort": true,
|
||||
"supports_xhigh_reasoning_effort": false,
|
||||
"supports_minimal_reasoning_effort": true
|
||||
},
|
||||
"gpt-5.2-codex": {
|
||||
"cache_read_input_token_cost": 1.75e-07,
|
||||
"cache_read_input_token_cost_priority": 3.5e-07,
|
||||
|
|
|
|||
|
|
@ -276,6 +276,8 @@ def test_gpt5_1_model_detection(gpt5_config: OpenAIGPT5Config):
|
|||
assert gpt5_config._supports_reasoning_effort_level("gpt-5.1", "none")
|
||||
assert gpt5_config._supports_reasoning_effort_level("gpt-5.1-2025-11-13", "none")
|
||||
assert gpt5_config._supports_reasoning_effort_level("gpt-5.1-chat-latest", "none")
|
||||
assert gpt5_config._supports_reasoning_effort_level("gpt-5.1-mini", "none")
|
||||
assert gpt5_config._supports_reasoning_effort_level("gpt-5.1-nano", "none")
|
||||
assert gpt5_config._supports_reasoning_effort_level("gpt-5.2", "none")
|
||||
assert gpt5_config._supports_reasoning_effort_level("gpt-5.2-2025-12-11", "none")
|
||||
# codex/pro/chat variants do not support none
|
||||
|
|
@ -304,6 +306,25 @@ def test_gpt5_1_temperature_with_reasoning_effort_none(config: OpenAIConfig):
|
|||
assert params["reasoning_effort"] == "none"
|
||||
|
||||
|
||||
def test_gpt5_1_mini_nano_temperature_with_reasoning_effort_none(
|
||||
config: OpenAIConfig,
|
||||
):
|
||||
"""Test that gpt-5.1-mini and gpt-5.1-nano support temperature when
|
||||
reasoning_effort='none' (registry parity with bare gpt-5.1)."""
|
||||
for model in ["gpt-5.1-mini", "gpt-5.1-nano"]:
|
||||
for temp in [0.0, 0.7, 1.0, 1.5]:
|
||||
params = config.map_openai_params(
|
||||
non_default_params={"temperature": temp, "reasoning_effort": "none"},
|
||||
optional_params={},
|
||||
model=model,
|
||||
drop_params=False,
|
||||
)
|
||||
assert params["temperature"] == temp, (
|
||||
f"{model} should preserve temperature={temp} with effort='none'"
|
||||
)
|
||||
assert params["reasoning_effort"] == "none"
|
||||
|
||||
|
||||
def test_gpt5_2_temperature_with_reasoning_effort_none(config: OpenAIConfig):
|
||||
"""Test that GPT-5.2 aligns with GPT-5.1 temperature rules when effort='none'."""
|
||||
for temp in [0.0, 0.3, 0.7, 1.0, 1.5]:
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue