From 755949952e05f10ae1f09f3faeae12fb1773dba1 Mon Sep 17 00:00:00 2001 From: mateo-berri <277851410+mateo-berri@users.noreply.github.com> Date: Thu, 29 Oct 2026 07:24:59 -0700 Subject: [PATCH 1/3] fix(fireworks): keep developer items in place for models that accept a later system message --- .../fireworks_ai/responses/transformation.py | 13 +++- ...odel_prices_and_context_window_backup.json | 35 +++++++++ model_prices_and_context_window.json | 35 +++++++++ ...t_fireworks_ai_responses_transformation.py | 76 +++++++++++++++++-- 4 files changed, 151 insertions(+), 8 deletions(-) diff --git a/litellm/llms/fireworks_ai/responses/transformation.py b/litellm/llms/fireworks_ai/responses/transformation.py index 86d9df9618a..0c2c83c2978 100644 --- a/litellm/llms/fireworks_ai/responses/transformation.py +++ b/litellm/llms/fireworks_ai/responses/transformation.py @@ -25,6 +25,7 @@ from litellm.utils import ( declared_mid_conversation_system_support, deployment_declared_mid_conversation_system_support, deployment_model_info, + get_bundled_model_cost_map, ) if TYPE_CHECKING: @@ -50,6 +51,12 @@ def _forwarded_user_params(litellm_params: GenericLiteLLMParams) -> Mapping[str, ) +def _bundled_cost_map_declares_mid_conversation_system(model: str) -> bool: + bundled_cost_map: Final[Mapping[str, object]] = get_bundled_model_cost_map() + entry: Final = bundled_cost_map.get(f"fireworks_ai/{model.removeprefix('fireworks_ai/')}") + return deployment_declared_mid_conversation_system_support(entry) is True + + class FireworksAIResponsesAPIConfig(OpenAIResponsesAPIConfig): @property def custom_llm_provider(self) -> LlmProviders: @@ -86,9 +93,13 @@ class FireworksAIResponsesAPIConfig(OpenAIResponsesAPIConfig): validated_input: Final = self._validate_input_param(input) model_info: Final = deployment_model_info(litellm_params) declared: Final = declared_mid_conversation_system_support(model, self.custom_llm_provider.value, model_info) + keeps_developer_items_in_place: Final = declared is True and ( + deployment_declared_mid_conversation_system_support(model_info) is True + or _bundled_cost_map_declares_mid_conversation_system(model) + ) folded: Final = ( None - if deployment_declared_mid_conversation_system_support(model_info) is True + if keeps_developer_items_in_place else instruction_items_folded( validated_input, response_api_optional_request_params, "user" if declared is False else "system" ) diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index 4fad4857c60..300d62faeed 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -27839,6 +27839,7 @@ "output_cost_per_token_priority": 7.2e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -28213,6 +28214,7 @@ "output_cost_per_token_priority": 7.2e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63404,6 +63406,7 @@ "output_cost_per_token_priority": 1.5e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -63423,6 +63426,7 @@ "output_cost_per_token_priority": 2.25e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -63465,6 +63469,7 @@ ], "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63502,6 +63507,7 @@ "output_cost_per_token_priority": 1.5e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -63521,6 +63527,7 @@ "output_cost_per_token_priority": 2.25e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -63595,6 +63602,7 @@ ], "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63615,6 +63623,7 @@ ], "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63638,6 +63647,7 @@ ], "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63694,6 +63704,7 @@ "output_cost_per_token_priority": 2.5e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63713,6 +63724,7 @@ "output_cost_per_token_priority": 3e-06, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63749,6 +63761,7 @@ "output_cost_per_token_priority": 6.375e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_vision": true }, @@ -63766,6 +63779,7 @@ "output_cost_per_token_priority": 2.5e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63785,6 +63799,7 @@ "output_cost_per_token_priority": 3e-06, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63855,6 +63870,7 @@ ], "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63878,6 +63894,7 @@ ], "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66671,6 +66688,7 @@ "output_cost_per_token_priority": 5.5e-06, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66689,6 +66707,7 @@ "output_cost_per_token_priority": 8.25e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66707,6 +66726,7 @@ "output_cost_per_token_priority": 5.5e-06, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66725,6 +66745,7 @@ "output_cost_per_token_priority": 8.25e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66740,6 +66761,7 @@ "output_cost_per_token": 6.6e-06, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66782,6 +66804,7 @@ "output_cost_per_token": 6.6e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66800,6 +66823,7 @@ "output_cost_per_token_priority": 6.25e-07, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_response_schema": true, "supports_tool_choice": true, "supports_vision": true @@ -66818,6 +66842,7 @@ "output_cost_per_token_priority": 9.375e-07, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_response_schema": true, "supports_tool_choice": true, "supports_vision": true @@ -66835,6 +66860,7 @@ "output_cost_per_token_priority": 6.25e-07, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_response_schema": true, "supports_tool_choice": true, "supports_vision": true @@ -66853,6 +66879,7 @@ "output_cost_per_token_priority": 9.375e-07, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_response_schema": true, "supports_tool_choice": true, "supports_vision": true @@ -66871,6 +66898,7 @@ "output_cost_per_token_priority": 4.05e-06, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_response_schema": true, "supports_tool_choice": true, "supports_vision": true @@ -73034,6 +73062,7 @@ "output_cost_per_token_priority": 1.875e-05, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -76525,6 +76554,7 @@ "output_cost_per_token_priority": 1.5e-06, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -76683,6 +76713,7 @@ "output_cost_per_token_priority": 9e-07, "source": "https://platform.minimax.io/docs/guides/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -80192,6 +80223,7 @@ ], "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -80208,6 +80240,7 @@ "output_cost_per_token": 3.08e-06, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -80224,6 +80257,7 @@ "output_cost_per_token": 8.4e-07, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -80240,6 +80274,7 @@ "output_cost_per_token": 8.4e-07, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index 4fad4857c60..300d62faeed 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -27839,6 +27839,7 @@ "output_cost_per_token_priority": 7.2e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -28213,6 +28214,7 @@ "output_cost_per_token_priority": 7.2e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63404,6 +63406,7 @@ "output_cost_per_token_priority": 1.5e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -63423,6 +63426,7 @@ "output_cost_per_token_priority": 2.25e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -63465,6 +63469,7 @@ ], "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63502,6 +63507,7 @@ "output_cost_per_token_priority": 1.5e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -63521,6 +63527,7 @@ "output_cost_per_token_priority": 2.25e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -63595,6 +63602,7 @@ ], "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63615,6 +63623,7 @@ ], "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63638,6 +63647,7 @@ ], "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63694,6 +63704,7 @@ "output_cost_per_token_priority": 2.5e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63713,6 +63724,7 @@ "output_cost_per_token_priority": 3e-06, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63749,6 +63761,7 @@ "output_cost_per_token_priority": 6.375e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_vision": true }, @@ -63766,6 +63779,7 @@ "output_cost_per_token_priority": 2.5e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63785,6 +63799,7 @@ "output_cost_per_token_priority": 3e-06, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63855,6 +63870,7 @@ ], "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63878,6 +63894,7 @@ ], "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66671,6 +66688,7 @@ "output_cost_per_token_priority": 5.5e-06, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66689,6 +66707,7 @@ "output_cost_per_token_priority": 8.25e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66707,6 +66726,7 @@ "output_cost_per_token_priority": 5.5e-06, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66725,6 +66745,7 @@ "output_cost_per_token_priority": 8.25e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66740,6 +66761,7 @@ "output_cost_per_token": 6.6e-06, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66782,6 +66804,7 @@ "output_cost_per_token": 6.6e-06, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -66800,6 +66823,7 @@ "output_cost_per_token_priority": 6.25e-07, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_response_schema": true, "supports_tool_choice": true, "supports_vision": true @@ -66818,6 +66842,7 @@ "output_cost_per_token_priority": 9.375e-07, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_response_schema": true, "supports_tool_choice": true, "supports_vision": true @@ -66835,6 +66860,7 @@ "output_cost_per_token_priority": 6.25e-07, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_response_schema": true, "supports_tool_choice": true, "supports_vision": true @@ -66853,6 +66879,7 @@ "output_cost_per_token_priority": 9.375e-07, "source": "https://docs.fireworks.ai/serverless/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_response_schema": true, "supports_tool_choice": true, "supports_vision": true @@ -66871,6 +66898,7 @@ "output_cost_per_token_priority": 4.05e-06, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_response_schema": true, "supports_tool_choice": true, "supports_vision": true @@ -73034,6 +73062,7 @@ "output_cost_per_token_priority": 1.875e-05, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -76525,6 +76554,7 @@ "output_cost_per_token_priority": 1.5e-06, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -76683,6 +76713,7 @@ "output_cost_per_token_priority": 9e-07, "source": "https://platform.minimax.io/docs/guides/pricing", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -80192,6 +80223,7 @@ ], "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -80208,6 +80240,7 @@ "output_cost_per_token": 3.08e-06, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -80224,6 +80257,7 @@ "output_cost_per_token": 8.4e-07, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, @@ -80240,6 +80274,7 @@ "output_cost_per_token": 8.4e-07, "source": "https://api.fireworks.ai/v1/serverless/models", "supports_function_calling": true, + "supports_mid_conversation_system": true, "supports_prompt_caching": true, "supports_reasoning": true, "supports_response_schema": true, diff --git a/tests/unit/llms/fireworks_ai/responses/test_fireworks_ai_responses_transformation.py b/tests/unit/llms/fireworks_ai/responses/test_fireworks_ai_responses_transformation.py index 7fc6ac2c436..aa231c978f8 100644 --- a/tests/unit/llms/fireworks_ai/responses/test_fireworks_ai_responses_transformation.py +++ b/tests/unit/llms/fireworks_ai/responses/test_fireworks_ai_responses_transformation.py @@ -172,10 +172,10 @@ def test_responses_call_forwards_previous_response_id_and_store() -> None: def test_responses_call_folds_developer_items_into_instructions() -> None: - client: Final = _mock_http_client(_fireworks_response("accounts/fireworks/models/kimi-k3")) + client: Final = _mock_http_client(_fireworks_response("accounts/fireworks/models/qwen3p8-2p4t-a95b")) with patch(HTTPX_CLIENT_FACTORY, return_value=client): litellm.responses( - model="fireworks_ai/accounts/fireworks/models/kimi-k3", + model="fireworks_ai/accounts/fireworks/models/qwen3p8-2p4t-a95b", input=[ {"role": "user", "content": "Hi there"}, {"role": "developer", "content": "Answer with exactly one word."}, @@ -214,6 +214,66 @@ def test_responses_call_keeps_developer_items_in_place_for_a_deployment_that_acc ) +@respx.mock +@pytest.mark.parametrize( + "model", + [ + "fireworks_ai/accounts/fireworks/models/kimi-k3", + "fireworks_ai/accounts/fireworks/models/glm-5p3", + "fireworks_ai/accounts/fireworks/routers/kimi-k3-fast", + "fireworks_ai/glm-5p3-us", + ], +) +def test_responses_call_keeps_developer_items_in_place_for_a_model_the_cost_map_declares_accepts_them( + model: str, +) -> None: + route: Final = _fireworks_responses_route(model.removeprefix("fireworks_ai/")) + litellm.responses( + model=model, + instructions="You are Codex", + input=[ + {"role": "user", "content": "Hi there"}, + {"role": "developer", "content": "Answer with exactly one word."}, + {"role": "user", "content": "What is the capital of France?"}, + ], + api_key="fw-test-key", + ) + body: Final = _routed_body(route) + assert body["instructions"] == "You are Codex" + assert tuple(body["input"]) == ( + {"role": "user", "content": "Hi there"}, + {"role": "system", "content": "Answer with exactly one word.", "type": "message"}, + {"role": "user", "content": "What is the capital of France?"}, + ) + + +@respx.mock +def test_responses_call_folds_developer_items_when_the_loaded_cost_map_turns_the_flag_off( + declare_mid_conversation_system: Callable[[str, bool], None], +) -> None: + declare_mid_conversation_system("fireworks_ai/accounts/fireworks/models/kimi-k3", False) + route: Final = _fireworks_responses_route("accounts/fireworks/models/kimi-k3") + litellm.responses( + model="fireworks_ai/accounts/fireworks/models/kimi-k3", + input=[ + {"role": "user", "content": "Hi there"}, + {"role": "developer", "content": "Answer with exactly one word."}, + {"role": "user", "content": "What is the capital of France?"}, + {"type": "message", "role": "assistant", "content": "Paris"}, + {"role": "developer", "content": "Reply with PONG."}, + ], + api_key="fw-test-key", + ) + body: Final = _routed_body(route) + assert body["instructions"] == "Answer with exactly one word." + assert tuple(body["input"]) == ( + {"role": "user", "content": "Hi there"}, + {"role": "user", "content": "What is the capital of France?"}, + {"type": "message", "role": "assistant", "content": "Paris"}, + {"role": "user", "content": "Reply with PONG.", "type": "message"}, + ) + + @respx.mock @pytest.mark.parametrize( ("model_info", "closing_role"), @@ -222,9 +282,9 @@ def test_responses_call_keeps_developer_items_in_place_for_a_deployment_that_acc def test_responses_call_sends_the_closing_developer_item_as_a_user_turn_only_for_a_declared_system_first_deployment( model_info: Mapping[str, object] | None, closing_role: str ) -> None: - route: Final = _fireworks_responses_route("accounts/fireworks/models/kimi-k3") + route: Final = _fireworks_responses_route("accounts/fireworks/models/qwen3p8-2p4t-a95b") litellm.responses( - model="fireworks_ai/accounts/fireworks/models/kimi-k3", + model="fireworks_ai/accounts/fireworks/models/qwen3p8-2p4t-a95b", input=[ {"role": "user", "content": "Hi there"}, {"role": "developer", "content": "Answer with exactly one word."}, @@ -402,7 +462,7 @@ def test_transform_request_forwards_non_string_instructions_and_input_untouched( developer_item: Final = {"role": "developer", "content": "Answer with exactly one word."} user_item: Final = {"role": "user", "content": "What is the capital of France?"} request: Final = FireworksAIResponsesAPIConfig().transform_responses_api_request( - model="accounts/fireworks/models/kimi-k3", + model="accounts/fireworks/models/qwen3p8-2p4t-a95b", input=cast(ResponseInputParam, [developer_item, user_item]), response_api_optional_request_params={"instructions": ["not", "a", "string"]}, litellm_params=GenericLiteLLMParams(), @@ -416,7 +476,7 @@ def test_transform_request_forwards_non_string_instructions_and_input_untouched( def test_responses_call_maps_pydantic_developer_items_and_replays_pydantic_output_items() -> None: - client: Final = _mock_http_client(_fireworks_response("accounts/fireworks/models/kimi-k3")) + client: Final = _mock_http_client(_fireworks_response("accounts/fireworks/models/qwen3p8-2p4t-a95b")) pydantic_input: Final = cast( ResponseInputParam, [ @@ -435,7 +495,9 @@ def test_responses_call_maps_pydantic_developer_items_and_replays_pydantic_outpu ) with patch(HTTPX_CLIENT_FACTORY, return_value=client): litellm.responses( - model="fireworks_ai/accounts/fireworks/models/kimi-k3", input=pydantic_input, api_key="fw-test-key" + model="fireworks_ai/accounts/fireworks/models/qwen3p8-2p4t-a95b", + input=pydantic_input, + api_key="fw-test-key", ) _, _, body = _sent_request(client) assert body["instructions"] == "Answer with exactly one word." From f1ea89e50401877001089ee33eefd00256fbf80e Mon Sep 17 00:00:00 2001 From: mateo-berri <277851410+mateo-berri@users.noreply.github.com> Date: Fri, 30 Oct 2026 23:47:49 -0700 Subject: [PATCH 2/3] fix(fireworks_ai): keep developer items in place for short Fireworks model names and stop flagging gpt-oss-120b --- litellm/llms/fireworks_ai/responses/transformation.py | 7 +++++-- litellm/model_prices_and_context_window_backup.json | 2 -- model_prices_and_context_window.json | 2 -- .../test_fireworks_ai_responses_transformation.py | 2 ++ 4 files changed, 7 insertions(+), 6 deletions(-) diff --git a/litellm/llms/fireworks_ai/responses/transformation.py b/litellm/llms/fireworks_ai/responses/transformation.py index 0c2c83c2978..636075e18b2 100644 --- a/litellm/llms/fireworks_ai/responses/transformation.py +++ b/litellm/llms/fireworks_ai/responses/transformation.py @@ -53,8 +53,11 @@ def _forwarded_user_params(litellm_params: GenericLiteLLMParams) -> Mapping[str, def _bundled_cost_map_declares_mid_conversation_system(model: str) -> bool: bundled_cost_map: Final[Mapping[str, object]] = get_bundled_model_cost_map() - entry: Final = bundled_cost_map.get(f"fireworks_ai/{model.removeprefix('fireworks_ai/')}") - return deployment_declared_mid_conversation_system_support(entry) is True + names: Final = (model.removeprefix("fireworks_ai/"), resolve_fireworks_resource_name(model)) + return any( + deployment_declared_mid_conversation_system_support(bundled_cost_map.get(f"fireworks_ai/{name}")) is True + for name in names + ) class FireworksAIResponsesAPIConfig(OpenAIResponsesAPIConfig): diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index 300d62faeed..480dfc542bd 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -27839,7 +27839,6 @@ "output_cost_per_token_priority": 7.2e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, - "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -28214,7 +28213,6 @@ "output_cost_per_token_priority": 7.2e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, - "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index 300d62faeed..480dfc542bd 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -27839,7 +27839,6 @@ "output_cost_per_token_priority": 7.2e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, - "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -28214,7 +28213,6 @@ "output_cost_per_token_priority": 7.2e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, - "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, diff --git a/tests/unit/llms/fireworks_ai/responses/test_fireworks_ai_responses_transformation.py b/tests/unit/llms/fireworks_ai/responses/test_fireworks_ai_responses_transformation.py index aa231c978f8..6538ccf1cac 100644 --- a/tests/unit/llms/fireworks_ai/responses/test_fireworks_ai_responses_transformation.py +++ b/tests/unit/llms/fireworks_ai/responses/test_fireworks_ai_responses_transformation.py @@ -222,6 +222,8 @@ def test_responses_call_keeps_developer_items_in_place_for_a_deployment_that_acc "fireworks_ai/accounts/fireworks/models/glm-5p3", "fireworks_ai/accounts/fireworks/routers/kimi-k3-fast", "fireworks_ai/glm-5p3-us", + "fireworks_ai/models/kimi-k3", + "fireworks_ai/routers/kimi-k3-fast", ], ) def test_responses_call_keeps_developer_items_in_place_for_a_model_the_cost_map_declares_accepts_them( From e2c91ed5ef55d3c7c27bf5df53a1f5360e5157cb Mon Sep 17 00:00:00 2001 From: mateo-berri <277851410+mateo-berri@users.noreply.github.com> Date: Sat, 31 Oct 2026 03:53:30 -0700 Subject: [PATCH 3/3] fix(fireworks): drop the mid-conversation system flag from nemotron-lightning nemotron-lightning-3p5-30b-a3b followed a later instruction less often with the developer item kept in place than folded, so both of its cost map keys go back to folding. --- litellm/model_prices_and_context_window_backup.json | 2 -- model_prices_and_context_window.json | 2 -- 2 files changed, 4 deletions(-) diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index 480dfc542bd..a6fa7a9ec31 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -63702,7 +63702,6 @@ "output_cost_per_token_priority": 2.5e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, - "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63777,7 +63776,6 @@ "output_cost_per_token_priority": 2.5e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, - "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index 480dfc542bd..a6fa7a9ec31 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -63702,7 +63702,6 @@ "output_cost_per_token_priority": 2.5e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, - "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true, @@ -63777,7 +63776,6 @@ "output_cost_per_token_priority": 2.5e-07, "source": "https://api.fireworks.ai/v1/serverless/models?format=nested", "supports_function_calling": true, - "supports_mid_conversation_system": true, "supports_reasoning": true, "supports_response_schema": true, "supports_tool_choice": true,