scriptoriumadmin/litellm
1
0
Fork 0
mirror of https://github.com/BerriAI/litellm.git synced 2026-09-09 22:31:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 87

fix(cost-map): stop advertising reasoning_effort max on the azure gpt-6-astra rows

Browse source 
Both Azure routes refuse it. A live call to the same deployment through
openai/deployments/gpt-6-astra/chat/completions on api-version 2025-04-01-preview
answers reasoning_effort max with a 400 unsupported_value naming none, low, medium,
high and xhigh as the values it takes, and xhigh returns 200, so azure/gpt-6-astra
and azure/us/gpt-6-astra now match the azure_ai row.
This commit is contained in:
mateo-berri 2026-09-05 22:31:32 -07:00
parent e8f311429e
commit e79f3ec520
4 changed files with 14 additions and 24 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
litellm/model_prices_and_context_window_backup.json
View file

@ -7237,7 +7237,7 @@
],
"supports_computer_use": true,
"supports_function_calling": true,
"supports_max_reasoning_effort": true,
"supports_max_reasoning_effort": false,
"supports_minimal_reasoning_effort": false,
"supports_native_streaming": true,
"supports_none_reasoning_effort": true,
@ -7503,7 +7503,7 @@
],
"supports_computer_use": true,
"supports_function_calling": true,
"supports_max_reasoning_effort": true,
"supports_max_reasoning_effort": false,
"supports_minimal_reasoning_effort": false,
"supports_native_streaming": true,
"supports_none_reasoning_effort": true,

4
litellm/router_utils/reasoning_effort_capability.py
View file

@ -10,8 +10,8 @@ opt-in. none is opt-out everywhere except the azure gpt-5 family, whose config r
UnsupportedParamsError without an explicit true.
xhigh is gated on the request path by the openai and azure gpt-5 configs. max is not gated there at
all: every entry carrying supports_max_reasoning_effort is Claude-family, and
anthropic/chat/transformation.py gates max on the output_config path while its reasoning_effort
all: outside the gpt-6-astra rows every entry carrying supports_max_reasoning_effort is Claude-family,
and anthropic/chat/transformation.py gates max on the output_config path while its reasoning_effort
path maps any level to a thinking budget. Making max opt-in is a deliberate trade, then, since an
explicit flag is the only signal that the tier is a real one rather than litellm rounding the level
to a budget, and a missing flag costs advisory metadata rather than a rejected request.

4
model_prices_and_context_window.json
View file

@ -7237,7 +7237,7 @@
],
"supports_computer_use": true,
"supports_function_calling": true,
"supports_max_reasoning_effort": true,
"supports_max_reasoning_effort": false,
"supports_minimal_reasoning_effort": false,
"supports_native_streaming": true,
"supports_none_reasoning_effort": true,
@ -7503,7 +7503,7 @@
],
"supports_computer_use": true,
"supports_function_calling": true,
"supports_max_reasoning_effort": true,
"supports_max_reasoning_effort": false,
"supports_minimal_reasoning_effort": false,
"supports_native_streaming": true,
"supports_none_reasoning_effort": true,

26
tests/test_litellm/router_utils/test_reasoning_effort_capability.py
View file

@ -394,30 +394,20 @@ class TestGpt6AstraAdvertisesItsDocumentedLevels:
[
("azure/gpt-6-astra", "azure"),
("azure/us/gpt-6-astra", "azure"),
("azure_ai/gpt-6-astra", "azure_ai"),
],
)
def test_a_foundry_deployment_also_advertises_none(self, local_model_cost_map, model, custom_llm_provider):
"""Microsoft Foundry serves the same model but its API accepts reasoning_effort none
(verified live: 200 with zero reasoning tokens, and it unlocks temperature), which
OpenAI's rejects, so an Azure deployment offers none on top of low through max."""
def test_an_azure_hosted_deployment_advertises_none_but_not_max(
self, local_model_cost_map, model, custom_llm_provider
):
"""Microsoft hosts the same model with a different level set than OpenAI does. Verified live
on both Azure routes: none returns 200 with zero reasoning tokens and unlocks temperature,
which OpenAI's API rejects, while max returns 400 unsupported_value naming none through
xhigh as the levels it does take."""
from litellm.utils import _get_model_info_helper
model_info = dict(_get_model_info_helper(model=model, custom_llm_provider=custom_llm_provider))
assert resolve_supported_reasoning_efforts(model_info, deployment_is_mapped=True) == (
"none",
"low",
"medium",
"high",
"xhigh",
"max",
)
def test_a_foundry_azure_ai_deployment_advertises_none_but_not_max(self, local_model_cost_map):
from litellm.utils import _get_model_info_helper
model_info = dict(_get_model_info_helper(model="azure_ai/gpt-6-astra", custom_llm_provider="azure_ai"))
assert resolve_supported_reasoning_efforts(model_info, deployment_is_mapped=True) == (
"none",
"low",