feat(model-profiles): distribute data across packages (#34024)

libs/partners/huggingface/langchain_huggingface/chat_models/huggingface.py

@@ -13,7 +13,11 @@ from langchain_core.callbacks.manager import (
     AsyncCallbackManagerForLLMRun,
     CallbackManagerForLLMRun,
 )
-from langchain_core.language_models import LanguageModelInput
+from langchain_core.language_models import (
+    LanguageModelInput,
+    ModelProfile,
+    ModelProfileRegistry,
+)
 from langchain_core.language_models.chat_models import (
     BaseChatModel,
     agenerate_from_stream,
@@ -61,9 +65,17 @@ from langchain_core.utils.pydantic import is_basemodel_subclass
 from pydantic import BaseModel, Field, model_validator
 from typing_extensions import Self
 
+from langchain_huggingface.data._profiles import _PROFILES
 from langchain_huggingface.llms.huggingface_endpoint import HuggingFaceEndpoint
 from langchain_huggingface.llms.huggingface_pipeline import HuggingFacePipeline
 
+_MODEL_PROFILES = cast("ModelProfileRegistry", _PROFILES)
+
+
+def _get_default_model_profile(model_name: str) -> ModelProfile:
+    default = _MODEL_PROFILES.get(model_name) or {}
+    return default.copy()
+
 
 @dataclass
 class TGI_RESPONSE:
@@ -580,6 +592,13 @@ class ChatHuggingFace(BaseChatModel):
             raise TypeError(msg)
         return self
 
+    @model_validator(mode="after")
+    def _set_model_profile(self) -> Self:
+        """Set model profile if not overridden."""
+        if self.profile is None and self.model_id:
+            self.profile = _get_default_model_profile(self.model_id)
+        return self
+
     def _create_chat_result(self, response: dict) -> ChatResult:
         generations = []
         token_usage = response.get("usage", {})

libs/partners/huggingface/langchain_huggingface/data/__init__.py

@@ -0,0 +1 @@
+"""Model profile data. All edits should be made in profile_augmentations.toml."""

libs/partners/huggingface/langchain_huggingface/data/_profiles.py

@@ -0,0 +1,187 @@
+"""Auto-generated model profiles.
+
+DO NOT EDIT THIS FILE MANUALLY.
+This file is generated by the langchain-profiles CLI tool.
+
+It contains data derived from the models.dev project.
+
+Source: https://github.com/sst/models.dev
+License: MIT License
+
+To update these data, refer to the instructions here:
+
+https://docs.langchain.com/oss/python/langchain/models#updating-or-overwriting-profile-data
+"""
+
+from typing import Any
+
+_PROFILES: dict[str, dict[str, Any]] = {
+    "moonshotai/Kimi-K2-Instruct": {
+        "max_input_tokens": 131072,
+        "max_output_tokens": 16384,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": False,
+        "tool_calling": True,
+    },
+    "moonshotai/Kimi-K2-Instruct-0905": {
+        "max_input_tokens": 262144,
+        "max_output_tokens": 16384,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": False,
+        "tool_calling": True,
+    },
+    "MiniMaxAI/MiniMax-M2": {
+        "max_input_tokens": 204800,
+        "max_output_tokens": 204800,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": True,
+        "tool_calling": True,
+    },
+    "Qwen/Qwen3-Embedding-8B": {
+        "max_input_tokens": 32000,
+        "max_output_tokens": 4096,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": False,
+        "tool_calling": False,
+    },
+    "Qwen/Qwen3-Embedding-4B": {
+        "max_input_tokens": 32000,
+        "max_output_tokens": 2048,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": False,
+        "tool_calling": False,
+    },
+    "Qwen/Qwen3-Coder-480B-A35B-Instruct": {
+        "max_input_tokens": 262144,
+        "max_output_tokens": 66536,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": False,
+        "tool_calling": True,
+    },
+    "Qwen/Qwen3-235B-A22B-Thinking-2507": {
+        "max_input_tokens": 262144,
+        "max_output_tokens": 131072,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": True,
+        "tool_calling": True,
+    },
+    "Qwen/Qwen3-Next-80B-A3B-Instruct": {
+        "max_input_tokens": 262144,
+        "max_output_tokens": 66536,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": False,
+        "tool_calling": True,
+    },
+    "Qwen/Qwen3-Next-80B-A3B-Thinking": {
+        "max_input_tokens": 262144,
+        "max_output_tokens": 131072,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": False,
+        "tool_calling": True,
+    },
+    "zai-org/GLM-4.5": {
+        "max_input_tokens": 131072,
+        "max_output_tokens": 98304,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": True,
+        "tool_calling": True,
+    },
+    "zai-org/GLM-4.6": {
+        "max_input_tokens": 200000,
+        "max_output_tokens": 128000,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": True,
+        "tool_calling": True,
+    },
+    "zai-org/GLM-4.5-Air": {
+        "max_input_tokens": 128000,
+        "max_output_tokens": 96000,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": True,
+        "tool_calling": True,
+    },
+    "deepseek-ai/Deepseek-V3-0324": {
+        "max_input_tokens": 16384,
+        "max_output_tokens": 8192,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": False,
+        "tool_calling": True,
+    },
+    "deepseek-ai/DeepSeek-R1-0528": {
+        "max_input_tokens": 163840,
+        "max_output_tokens": 163840,
+        "image_inputs": False,
+        "audio_inputs": False,
+        "video_inputs": False,
+        "image_outputs": False,
+        "audio_outputs": False,
+        "video_outputs": False,
+        "reasoning_output": True,
+        "tool_calling": True,
+    },
+}