7a4594b682
Closes #37042 --- `AnthropicPromptCachingMiddleware` was unconditionally setting top-level `cache_control` in `model_settings` for any `ChatAnthropic` subclass. That field is direct-Anthropic-API only — `ChatAnthropicBedrock` (which subclasses `ChatAnthropic` and passed the existing `isinstance` gate) errored with `cache_control: Extra inputs are not permitted`. Investigating that surfaced a related regression: PR #35967 also deleted the block-level `cache_control` injection in `_get_request_payload`, which silently disabled caching entirely for non-direct subclasses (Bedrock had been falling back to in-block breakpoints). This restores both paths. ## Changes - Add `_is_direct_anthropic_llm_type` predicate that allowlists `_llm_type == "anthropic-chat"`. Both the middleware's `_supports_automatic_caching` and the new branch in `ChatAnthropic._get_request_payload` route through it, so any subclass that overrides `_llm_type` (Bedrock today, future direct-API variants tomorrow) is treated as non-direct by default. Replaces the prior substring-matching denylist on `"bedrock"`/`"vertex"`. - Restore `_collect_code_execution_tool_ids`, `_is_code_execution_related_block`, and a new `_apply_cache_control_to_last_eligible_block` helper in `chat_models`. For non-direct subclasses, `_get_request_payload` now pops `cache_control` from kwargs and walks messages newest-to-oldest, attaching the breakpoint to the last block that isn't `code_execution`-related (Anthropic forbids breakpoints on those). - Emit `UserWarning` when `cache_control` is requested but every candidate block is `code_execution`-related — previously a silent drop. - `AnthropicPromptCachingMiddleware._apply_caching` now sets the top-level `cache_control` only when `_supports_automatic_caching(request.model)`. System-message and tool-definition breakpoints continue to apply for all `ChatAnthropic` subclasses, since those are accepted by every transport. - Note: `ChatAnthropicVertex` does not subclass `ChatAnthropic` (it lives in `langchain-google-vertexai` and ships its own `_get_request_payload`), so the chat-models changes here only affect Bedrock. The middleware-side gate covers Vertex implicitly via the `isinstance(request.model, ChatAnthropic)` check that already excludes it.
🦜🪪 langchain-model-profiles
Warning
This package is currently in development and the API is subject to change.
CLI tool for updating model profile data in LangChain integration packages.
Quick Install
pip install langchain-model-profiles
🤔 What is this?
langchain-model-profiles is a CLI tool for fetching and updating model capability data from models.dev for use in LangChain integration packages.
LangChain chat models expose a .profile field that provides programmatic access to model capabilities such as context window sizes, supported modalities, tool calling, structured output, and more. This CLI tool helps maintainers keep that data up-to-date.
Data sources
This package is built on top of the excellent work by the models.dev project, an open source initiative that provides model capability data.
LangChain model profiles augment the data from models.dev with some additional fields. We intend to keep this aligned with the upstream project as it evolves.
📖 Documentation
For full documentation, see the API reference. For conceptual guides, tutorials, and examples on using LangChain, see the LangChain Docs. You can also chat with the docs using Chat LangChain.
Usage
Update model profile data for a specific provider:
langchain-profiles refresh --provider anthropic --data-dir ./langchain_anthropic/data
This downloads the latest model data from models.dev, merges it with any augmentations defined in profile_augmentations.toml, and generates a profiles.py file.