Submission checklist

• This is a feature request, not a bug report or usage question.
• I added a clear and descriptive title that summarizes the feature request.
• I used the GitHub search to find a similar feature request and didn't find it.
• I checked the LangChain documentation and API reference to see if this feature already exists.
• This is not related to the langchain-community package.

Package (Required)

• langchain
• langchain-openai
• langchain-anthropic
• langchain-classic
• langchain-core
• langchain-model-profiles
• langchain-tests
• langchain-text-splitters
• langchain-chroma
• langchain-deepseek
• langchain-exa
• langchain-fireworks
• langchain-groq
• langchain-huggingface
• langchain-mistralai
• langchain-nomic
• langchain-ollama
• langchain-openrouter
• langchain-perplexity
• langchain-qdrant
• langchain-xai
• Other / not sure / general

Feature Description

The Feb 2026 changelog states that SummarizationMiddleware auto-triggers when a chat model raises ContextOverflowError. Today only langchain-anthropic and langchain-openai map provider context-limit errors to that exception.

ChatGroq raises groq.BadRequestError with code context_length_exceeded but never converts it to ContextOverflowError, so summarization does not auto-trigger for Groq users.

ChatMistralAI surfaces context overflow as httpx.HTTPStatusError (HTTP 400) without conversion, so the same gap exists for Mistral.

Additionally, BaseChatModel.get_num_tokens_from_messages() ignores tool schemas (with a warning), which underestimates tokens for tool-heavy agents and can delay summarization.

Use Case

I use SummarizationMiddleware with Groq and Mistral models in agents with many tools. When context limits are hit, I get raw provider errors instead of ContextOverflowError, so automatic summarization never runs. Token estimates also omit tool schemas, so summarization triggers too late.

Proposed Solution

1. langchain-groq: Add GroqContextOverflowError (subclasses groq.BadRequestError and ContextOverflowError), map overflow in _generate/_agenerate/_stream/_astream. Override get_num_tokens_from_messages to include tool schemas via tiktoken cl100k_base.

2. langchain-mistralai: Same pattern with MistralContextOverflowError (httpx.HTTPStatusError + ContextOverflowError).

3. langchain-tests: Add raises_on_context_overflow opt-in property (default False) and integration tests.

Draft PRs (closed pending issue link + assignee):

• feat(groq): raise ContextOverflowError on context limit exceeded #37529
• feat(mistralai): raise ContextOverflowError on context limit exceeded #37530
• feat(standard-tests): add ContextOverflowError behavior test #37531

Alternatives Considered

Manual try/except around every model call to detect context overflow strings — fragile and duplicates logic already in Anthropic/OpenAI integrations.

Additional Context

AI-assisted contribution. Happy to split into separate issues/PRs if maintainers prefer.