diff --git a/litellm/__init__.py b/litellm/__init__.py index 09585bda7bf..fcfc4768ff3 100644 --- a/litellm/__init__.py +++ b/litellm/__init__.py @@ -1700,6 +1700,9 @@ if TYPE_CHECKING: from .llms.vertex_ai.vertex_ai_partner_models.ai21.transformation import ( VertexAIAi21Config as VertexAIAi21Config, ) + from .llms.vertex_ai.vertex_ai_partner_models.mistral.transformation import ( + VertexAIMistralConfig as VertexAIMistralConfig, + ) from .llms.bedrock.chat.invoke_handler import ( AmazonCohereChatConfig as AmazonCohereChatConfig, ) diff --git a/litellm/_lazy_imports_registry.py b/litellm/_lazy_imports_registry.py index 4c478b51ed1..9cfcb9e41f7 100644 --- a/litellm/_lazy_imports_registry.py +++ b/litellm/_lazy_imports_registry.py @@ -184,6 +184,7 @@ LLM_CONFIG_NAMES: Final = ( "VertexAIAnthropicConfig", "VertexAILlama3Config", "VertexAIAi21Config", + "VertexAIMistralConfig", "AmazonCohereChatConfig", "AmazonBedrockGlobalConfig", "AmazonAI21Config", @@ -771,6 +772,10 @@ _LLM_CONFIGS_IMPORT_MAP: Final = { ".llms.vertex_ai.vertex_ai_partner_models.ai21.transformation", "VertexAIAi21Config", ), + "VertexAIMistralConfig": ( + ".llms.vertex_ai.vertex_ai_partner_models.mistral.transformation", + "VertexAIMistralConfig", + ), "AmazonCohereChatConfig": ( ".llms.bedrock.chat.invoke_handler", "AmazonCohereChatConfig", diff --git a/litellm/litellm_core_utils/get_supported_openai_params.py b/litellm/litellm_core_utils/get_supported_openai_params.py index 915a03025d9..08b8816e17d 100644 --- a/litellm/litellm_core_utils/get_supported_openai_params.py +++ b/litellm/litellm_core_utils/get_supported_openai_params.py @@ -190,7 +190,7 @@ def get_supported_openai_params( elif custom_llm_provider == "vertex_ai" or custom_llm_provider == "vertex_ai_beta": if request_type == "chat_completion": if model.startswith("mistral"): - return litellm.MistralConfig().get_supported_openai_params(model=model) + return litellm.VertexAIMistralConfig().get_supported_openai_params(model=model) elif model.startswith("codestral"): return litellm.CodestralTextCompletionConfig().get_supported_openai_params(model=model) elif model.startswith("claude"): diff --git a/litellm/llms/mistral/chat/transformation.py b/litellm/llms/mistral/chat/transformation.py index a76a8a3e98c..f77e828b59a 100644 --- a/litellm/llms/mistral/chat/transformation.py +++ b/litellm/llms/mistral/chat/transformation.py @@ -11,6 +11,7 @@ from typing import TYPE_CHECKING, Any, Final, Literal, cast, get_type_hints, ove import httpx +from litellm._logging import verbose_logger from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj from litellm.litellm_core_utils.prompt_templates.common_utils import ( handle_messages_with_content_list_to_str_conversion, @@ -20,16 +21,37 @@ from litellm.llms.openai.chat.gpt_transformation import ( OpenAIChatCompletionStreamingHandler, OpenAIGPTConfig, ) +from litellm.router_utils.reasoning_effort_capability import ( + declared_reasoning_efforts_for_model, + nearest_declared_reasoning_effort, +) from litellm.secret_managers.main import get_secret_str from litellm.types.llms.mistral import MistralThinkingBlock, MistralToolCallMessage from litellm.types.llms.openai import AllMessageValues from litellm.types.utils import ModelResponse, ModelResponseStream -from litellm.utils import convert_to_model_response_object +from litellm.utils import convert_to_model_response_object, supports_reasoning if TYPE_CHECKING: import tiktoken +def _accepted_reasoning_effort(model: str, requested: str, custom_llm_provider: str) -> str: + declared: Final = declared_reasoning_efforts_for_model(model, custom_llm_provider) + if declared is None: + return requested + accepted: Final = nearest_declared_reasoning_effort(requested, declared) + if accepted != requested: + verbose_logger.debug( + "%s: %s takes reasoning_effort %s, sending %s in place of %s", + custom_llm_provider, + model, + declared, + accepted, + requested, + ) + return accepted + + class MistralConfig(OpenAIGPTConfig): """ Reference: https://docs.mistral.ai/api/ @@ -86,8 +108,16 @@ class MistralConfig(OpenAIGPTConfig): def get_config(cls): return super().get_config() + @property + def custom_llm_provider(self) -> str: + return "mistral" + def get_supported_openai_params(self, model: str) -> list[str]: - supported_params: Final = [ + is_magistral: Final = "magistral" in model.lower() + accepts_reasoning_effort: Final = is_magistral or supports_reasoning( + model=model, custom_llm_provider=self.custom_llm_provider + ) + return [ "stream", "temperature", "top_p", @@ -99,14 +129,10 @@ class MistralConfig(OpenAIGPTConfig): "stop", "response_format", "parallel_tool_calls", + *(("thinking",) if is_magistral else ()), + *(("reasoning_effort",) if accepts_reasoning_effort else ()), ] - # Add reasoning support for magistral models - if "magistral" in model.lower(): - supported_params.extend(["thinking", "reasoning_effort"]) - - return supported_params - def _map_tool_choice(self, tool_choice: str) -> str: if tool_choice == "auto" or tool_choice == "none": return tool_choice @@ -171,10 +197,9 @@ class MistralConfig(OpenAIGPTConfig): optional_params["extra_body"] = {"random_seed": value} if param == "response_format": optional_params["response_format"] = value - if param == "reasoning_effort" and "magistral" in model.lower(): - # Flag that we need to add reasoning system prompt - optional_params["_add_reasoning_prompt"] = True - if param == "thinking" and "magistral" in model.lower(): + if param == "reasoning_effort" and "magistral" not in model.lower(): + optional_params["reasoning_effort"] = _accepted_reasoning_effort(model, value, self.custom_llm_provider) + if param in ("reasoning_effort", "thinking") and "magistral" in model.lower(): # Flag that we need to add reasoning system prompt optional_params["_add_reasoning_prompt"] = True if param == "parallel_tool_calls": @@ -534,11 +559,13 @@ class MistralConfig(OpenAIGPTConfig): if "magistral" in model.lower() and optional_params.get("_add_reasoning_prompt", False): messages = self._add_reasoning_system_prompt_if_needed(messages, optional_params) + upstream_params: Final = {key: value for key, value in optional_params.items() if key != "client_metadata"} + # Call parent transform_request which handles _transform_messages return super().transform_request( model=model, messages=messages, - optional_params=optional_params, + optional_params=upstream_params, litellm_params=litellm_params, headers=headers, ) diff --git a/litellm/llms/vertex_ai/vertex_ai_partner_models/mistral/transformation.py b/litellm/llms/vertex_ai/vertex_ai_partner_models/mistral/transformation.py new file mode 100644 index 00000000000..18321c8768a --- /dev/null +++ b/litellm/llms/vertex_ai/vertex_ai_partner_models/mistral/transformation.py @@ -0,0 +1,7 @@ +from litellm.llms.mistral.chat.transformation import MistralConfig + + +class VertexAIMistralConfig(MistralConfig): + @property + def custom_llm_provider(self) -> str: + return "vertex_ai" diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index 377e30a23b7..e0c726b1e2c 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -37097,6 +37097,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37178,6 +37182,15 @@ "max_tokens": 131072, "mode": "chat", "output_cost_per_token": 4.4e-06, + "reasoning_effort_levels": [ + "none", + "minimal", + "low", + "medium", + "high", + "xhigh", + "max" + ], "source": "https://docs.mistral.ai/models/zai-glm-5-2", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37195,6 +37208,11 @@ "max_tokens": 131072, "mode": "chat", "output_cost_per_token": 4.4e-06, + "reasoning_effort_levels": [ + "low", + "high", + "max" + ], "source": "https://docs.mistral.ai/models/zai-glm-5-3", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37212,6 +37230,11 @@ "max_tokens": 131072, "mode": "chat", "output_cost_per_token": 4.4e-06, + "reasoning_effort_levels": [ + "low", + "high", + "max" + ], "source": "https://docs.mistral.ai/models/zai-glm-5-3", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37229,6 +37252,11 @@ "max_tokens": 131072, "mode": "chat", "output_cost_per_token": 4.4e-06, + "reasoning_effort_levels": [ + "low", + "high", + "max" + ], "source": "https://docs.mistral.ai/models/zai-glm-5-3", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37246,6 +37274,15 @@ "max_tokens": 131072, "mode": "chat", "output_cost_per_token": 4.4e-06, + "reasoning_effort_levels": [ + "none", + "minimal", + "low", + "medium", + "high", + "xhigh", + "max" + ], "source": "https://docs.mistral.ai/models/zai-glm-5-2", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37541,6 +37578,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37601,6 +37642,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37618,6 +37663,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37651,6 +37700,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37682,6 +37735,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 6e-07, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-small-4-0-26-03", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -60027,6 +60084,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 6e-07, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-small-4-0-26-03", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -63352,6 +63413,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -63369,6 +63434,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -63386,6 +63455,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -63403,6 +63476,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 6e-07, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-small-4-0-26-03", "supports_assistant_prefill": true, "supports_function_calling": true, diff --git a/litellm/router_utils/reasoning_effort_capability.py b/litellm/router_utils/reasoning_effort_capability.py index 7b145c15a07..1d7656f253e 100644 --- a/litellm/router_utils/reasoning_effort_capability.py +++ b/litellm/router_utils/reasoning_effort_capability.py @@ -103,6 +103,25 @@ def declared_reasoning_efforts_for_model(model: str, custom_llm_provider: str) - return declared_reasoning_efforts(entry) +REASONING_EFFORT_STRENGTH_ORDER: Final = ("minimal", "low", "medium", "high", "xhigh", "max") +_STRENGTH_RANK: Final = MappingProxyType({effort: rank for rank, effort in enumerate(REASONING_EFFORT_STRENGTH_ORDER)}) + + +def nearest_declared_reasoning_effort(requested: str, declared: Sequence[str]) -> str: + """Rounds a request up to the weakest declared level at least as strong as it, and down to the + strongest declared level when it asks for more than the model has, so the caller gets no less + reasoning than it asked for instead of a rejected call. none is the off switch rather than a + strength, so it is never rounded onto the ladder and no level is rounded down to it: a caller + who turned reasoning off must not be billed for it, and a model that cannot turn it off says so + itself. A level outside the strength order is likewise returned as is for upstream to judge.""" + ranked: Final = sorted( + (effort for effort in declared if effort in _STRENGTH_RANK), key=lambda effort: _STRENGTH_RANK[effort] + ) + if requested in ranked or requested not in _STRENGTH_RANK or not ranked: + return requested + return next((effort for effort in ranked if _STRENGTH_RANK[effort] >= _STRENGTH_RANK[requested]), ranked[-1]) + + def _supports_none_reasoning_effort(model_info: Mapping[str, object], flag: object) -> bool: """Opt-in only where a request path refuses the level. AzureOpenAIGPT5Config raises UnsupportedParamsError on reasoning_effort='none' without an explicit true, and it is selected diff --git a/litellm/utils.py b/litellm/utils.py index ec11895a8b5..f2315651a53 100644 --- a/litellm/utils.py +++ b/litellm/utils.py @@ -4526,7 +4526,7 @@ def get_optional_params( drop_params=bool(drop_params), ) else: - optional_params = litellm.MistralConfig().map_openai_params( + optional_params = litellm.VertexAIMistralConfig().map_openai_params( model=model, non_default_params=non_default_params, optional_params=optional_params, @@ -8396,7 +8396,7 @@ class ProviderConfigManager: elif model in litellm.vertex_mistral_models: if "codestral" in model: return litellm.CodestralTextCompletionConfig() - return litellm.MistralConfig() + return litellm.VertexAIMistralConfig() elif model in litellm.vertex_ai_ai21_models: return litellm.VertexAIAi21Config() else: diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index 377e30a23b7..e0c726b1e2c 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -37097,6 +37097,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37178,6 +37182,15 @@ "max_tokens": 131072, "mode": "chat", "output_cost_per_token": 4.4e-06, + "reasoning_effort_levels": [ + "none", + "minimal", + "low", + "medium", + "high", + "xhigh", + "max" + ], "source": "https://docs.mistral.ai/models/zai-glm-5-2", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37195,6 +37208,11 @@ "max_tokens": 131072, "mode": "chat", "output_cost_per_token": 4.4e-06, + "reasoning_effort_levels": [ + "low", + "high", + "max" + ], "source": "https://docs.mistral.ai/models/zai-glm-5-3", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37212,6 +37230,11 @@ "max_tokens": 131072, "mode": "chat", "output_cost_per_token": 4.4e-06, + "reasoning_effort_levels": [ + "low", + "high", + "max" + ], "source": "https://docs.mistral.ai/models/zai-glm-5-3", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37229,6 +37252,11 @@ "max_tokens": 131072, "mode": "chat", "output_cost_per_token": 4.4e-06, + "reasoning_effort_levels": [ + "low", + "high", + "max" + ], "source": "https://docs.mistral.ai/models/zai-glm-5-3", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37246,6 +37274,15 @@ "max_tokens": 131072, "mode": "chat", "output_cost_per_token": 4.4e-06, + "reasoning_effort_levels": [ + "none", + "minimal", + "low", + "medium", + "high", + "xhigh", + "max" + ], "source": "https://docs.mistral.ai/models/zai-glm-5-2", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37541,6 +37578,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37601,6 +37642,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37618,6 +37663,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37651,6 +37700,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -37682,6 +37735,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 6e-07, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-small-4-0-26-03", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -60027,6 +60084,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 6e-07, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-small-4-0-26-03", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -63352,6 +63413,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -63369,6 +63434,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -63386,6 +63455,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 7.5e-06, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-medium-3-5-26-04", "supports_assistant_prefill": true, "supports_function_calling": true, @@ -63403,6 +63476,10 @@ "max_tokens": 262144, "mode": "chat", "output_cost_per_token": 6e-07, + "reasoning_effort_levels": [ + "none", + "high" + ], "source": "https://docs.mistral.ai/models/model-cards/mistral-small-4-0-26-03", "supports_assistant_prefill": true, "supports_function_calling": true, diff --git a/tests/test_litellm/llms/mistral/test_mistral_chat_transformation.py b/tests/test_litellm/llms/mistral/test_mistral_chat_transformation.py index 15694d9f218..8fb3b3c43df 100644 --- a/tests/test_litellm/llms/mistral/test_mistral_chat_transformation.py +++ b/tests/test_litellm/llms/mistral/test_mistral_chat_transformation.py @@ -51,14 +51,19 @@ class TestMistralReasoningSupport: assert "reasoning_effort" in supported_params assert "thinking" in supported_params - # Test non-magistral model doesn't include reasoning parameters + supported_params_reasoning = mistral_config.get_supported_openai_params( + "mistral/mistral-medium-latest" + ) + assert "reasoning_effort" in supported_params_reasoning + assert "thinking" not in supported_params_reasoning + supported_params_normal = mistral_config.get_supported_openai_params( "mistral/mistral-large-latest" ) assert "reasoning_effort" not in supported_params_normal assert "thinking" not in supported_params_normal - def test_map_openai_params_reasoning_effort(self): + def test_map_openai_params_reasoning_effort(self, local_model_cost_map): """Test that reasoning_effort parameter is properly mapped for magistral models.""" mistral_config = MistralConfig() @@ -73,16 +78,93 @@ class TestMistralReasoningSupport: assert result.get("_add_reasoning_prompt") is True - # Test reasoning_effort ignored for non-magistral model optional_params_normal = {} result_normal = mistral_config.map_openai_params( non_default_params={"reasoning_effort": "low"}, optional_params=optional_params_normal, - model="mistral/mistral-large-latest", + model="mistral/mistral-medium-latest", drop_params=False, ) assert "_add_reasoning_prompt" not in result_normal + assert result_normal["reasoning_effort"] == "high" + + @pytest.mark.parametrize( + ("model", "requested", "sent"), + [ + ("mistral-medium-latest", "high", "high"), + ("mistral-medium-latest", "none", "none"), + ("mistral-medium-latest", "low", "high"), + ("mistral-medium-latest", "medium", "high"), + ("mistral-medium-latest", "xhigh", "high"), + ("mistral-small-latest", "medium", "high"), + ("mistral-vibe-cli-latest", "medium", "high"), + ("zai-glm-5", "none", "none"), + ("zai-glm-5", "minimal", "low"), + ("zai-glm-5", "medium", "high"), + ("zai-glm-5", "xhigh", "max"), + ("zai-glm-5-2", "medium", "medium"), + ("zai-glm-5-2", "xhigh", "xhigh"), + ], + ) + def test_reasoning_effort_is_sent_as_a_level_the_model_accepts(self, local_model_cost_map, model, requested, sent): + import litellm + + optional_params = litellm.get_optional_params( + model=model, + custom_llm_provider="mistral", + reasoning_effort=requested, + ) + assert optional_params["reasoning_effort"] == sent + + def test_reasoning_effort_is_forwarded_verbatim_when_the_map_declares_no_levels( + self, local_model_cost_map, monkeypatch + ): + import litellm + + monkeypatch.setitem( + litellm.model_cost, + "mistral/undeclared-reasoner", + {"litellm_provider": "mistral", "mode": "chat", "supports_reasoning": True}, + ) + optional_params = litellm.get_optional_params( + model="undeclared-reasoner", + custom_llm_provider="mistral", + reasoning_effort="medium", + ) + assert optional_params["reasoning_effort"] == "medium" + + def test_reasoning_effort_stays_unsupported_for_non_reasoning_models(self): + import litellm + + with pytest.raises(litellm.UnsupportedParamsError): + litellm.get_optional_params( + model="codestral-latest", + custom_llm_provider="mistral", + reasoning_effort="high", + ) + + dropped = litellm.get_optional_params( + model="codestral-latest", + custom_llm_provider="mistral", + reasoning_effort="high", + drop_params=True, + ) + assert "reasoning_effort" not in dropped + + def test_client_metadata_stripped_from_request(self): + mistral_config = MistralConfig() + + request = mistral_config.transform_request( + model="mistral-medium-latest", + messages=[{"role": "user", "content": "hi"}], + optional_params={"client_metadata": {"originator": "codex_cli_rs"}, "temperature": 0.2}, + litellm_params={}, + headers={}, + ) + + assert "client_metadata" not in request + assert request["temperature"] == 0.2 def test_map_openai_params_thinking(self): """Test that thinking parameter is properly mapped for magistral models.""" diff --git a/tests/test_litellm/llms/vertex_ai/vertex_ai_partner_models/mistral/test_vertex_ai_partner_models_mistral_transformation.py b/tests/test_litellm/llms/vertex_ai/vertex_ai_partner_models/mistral/test_vertex_ai_partner_models_mistral_transformation.py new file mode 100644 index 00000000000..f7df4507651 --- /dev/null +++ b/tests/test_litellm/llms/vertex_ai/vertex_ai_partner_models/mistral/test_vertex_ai_partner_models_mistral_transformation.py @@ -0,0 +1,17 @@ +import litellm + + +def test_reasoning_effort_stays_unsupported_on_vertex_partner_models(local_model_cost_map): + assert "reasoning_effort" in litellm.get_supported_openai_params( + model="mistral-medium-3", custom_llm_provider="mistral" + ) + assert "reasoning_effort" not in litellm.get_supported_openai_params( + model="mistral-medium-3", custom_llm_provider="vertex_ai" + ) + dropped = litellm.get_optional_params( + model="mistral-medium-3", + custom_llm_provider="vertex_ai", + reasoning_effort="high", + drop_params=True, + ) + assert "reasoning_effort" not in dropped diff --git a/tests/test_litellm/router_utils/test_reasoning_effort_capability.py b/tests/test_litellm/router_utils/test_reasoning_effort_capability.py index ccd6766b13a..adee44aa8a3 100644 --- a/tests/test_litellm/router_utils/test_reasoning_effort_capability.py +++ b/tests/test_litellm/router_utils/test_reasoning_effort_capability.py @@ -4,6 +4,7 @@ import litellm from litellm.router_utils.reasoning_effort_capability import ( deployment_is_catalog_mapped, intersect_supported_reasoning_efforts, + nearest_declared_reasoning_effort, resolve_supported_reasoning_efforts, ) @@ -415,3 +416,26 @@ class TestGpt6AstraAdvertisesItsDocumentedLevels: "high", "xhigh", ) + + +class TestNearestDeclaredReasoningEffort: + def test_a_declared_level_is_kept(self): + assert nearest_declared_reasoning_effort("high", ("none", "high")) == "high" + assert nearest_declared_reasoning_effort("none", ("none", "high")) == "none" + + def test_an_undeclared_level_rounds_up_to_the_next_declared_one(self): + assert nearest_declared_reasoning_effort("medium", ("none", "high")) == "high" + assert nearest_declared_reasoning_effort("minimal", ("low", "high", "max")) == "low" + assert nearest_declared_reasoning_effort("xhigh", ("low", "high", "max")) == "max" + + def test_none_is_a_switch_that_is_never_rounded_in_either_direction(self): + assert nearest_declared_reasoning_effort("none", ("low", "high", "max")) == "none" + assert nearest_declared_reasoning_effort("medium", ("none",)) == "medium" + + def test_a_level_above_the_ceiling_takes_the_strongest_declared_one(self): + assert nearest_declared_reasoning_effort("max", ("none", "high")) == "high" + assert nearest_declared_reasoning_effort("xhigh", ("none", "low", "medium", "high")) == "high" + + def test_a_level_outside_the_strength_order_is_left_for_upstream(self): + assert nearest_declared_reasoning_effort("turbo", ("none", "high")) == "turbo" + assert nearest_declared_reasoning_effort("medium", ()) == "medium"