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

I’d like to propose an IntentAwareModelRouterMiddleware for langchain_v1 agents that routes requests across model tiers (fast / balanced / quality) based on prompt intent.

Motivation: Agent apps often need to trade off latency, cost, and output quality. Today this is usually hardcoded per app. A built-in middleware would provide a reusable default policy while still allowing custom policies.

Proposed behavior:

• Route to quality model for high-stakes intent (e.g., legal/medical/financial/compliance/security-sensitive prompts)
• Route to fast model for simple short tasks (e.g., summarize/rewrite/extract/classify)
• Route to balanced model for everything else
• Keep ModelFallbackMiddleware compatibility for failure recovery
• Add routing trace metadata to responses for observability:
  • policy
  • policy_version
  • route_reason
  • decision_features

Scope:

• New middleware in langchain_v1 middleware package
• Public export in middleware __init__
• Unit tests for:
  • route selection (fast / balanced / quality)
  • metadata emission
  • composition with fallback middleware

Backward compatibility: No breaking API changes. Middleware is opt-in.

Prior art / implementation status: I already have a working implementation + tests in a PR: https://github.com/langchain-ai/langchain/pull/37009

Happy to adjust naming, policy defaults, or scope based on maintainer feedback.

Use Case

I’m building agent apps where some requests are high-stakes and others are simple transforms. Using one fixed model wastes cost or hurts quality. This feature gives a reusable, transparent middleware policy for model selection with fallback compatibility.

Proposed Solution

1. Add IntentAwareModelRouterMiddleware in langchain_v1 middleware.
2. Policy: route to quality for high-stakes prompts, fast for simple short tasks, balanced otherwise.
3. Attach routing trace metadata (policy, policy_version, route_reason, decision_features).
4. Keep compatibility with ModelFallbackMiddleware.

Alternatives Considered

1. Per-application custom routing logic.
2. Only fallback/retry without pre-routing.
3. Static model selection in config.

Additional Context

Related PR: https://github.com/langchain-ai/langchain/pull/37009