[FEATURE] Intelligent Auto-Model Selection (Prompt Routing) for Claude Chat & CLI

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

As a Claude Pro subscriber, managing token consumption and switching between models manually is a constant challenge. Currently, every prompt in a session goes through the selected high-tier model (like Opus or Sonnet). When asking simple, routine questions or doing quick text/code edits, it drains the session and usage limits unnecessarily. There is no native way for the system to automatically optimize compute resources based on prompt complexity.

Proposed Solution

I propose integrating an "Auto" model selection option (Intelligent Prompt Routing) directly into the interface. A lightweight model (like Haiku) could analyze the incoming prompt and automatically route it to the appropriate model based on complexity:

• Haiku: For simple explanations, quick text/code editing, or short queries.
• Sonnet: For day-to-day coding, debugging, and standard analysis.
• Opus: Only when deep reasoning, advanced math, or complex architecture is required.

This would significantly improve user experience, protect user session limits from being drained by simple questions, and optimize server load on Anthropic's end. This pattern is already proven successful in services like Gemini Auto.

Alternative Solutions

_No response_

Priority

High - Significant impact on productivity

Feature Category

Performance and speed

Use Case Example

Example scenario:

1. I am inside a Claude Project containing large knowledge files (like extensive codebase documentation or API schemas).
2. Inside this Project, I open a new chat to ask a simple conceptual question, such as: "What is the general purpose of the second library listed in the files?"
3. Currently, because I am inside the project, this simple query is processed by a heavy model (Sonnet/Opus), draining my session limits heavily just to read a short answer.
4. With this feature, Haiku could automatically handle these simple exploratory or summarization prompts within the Project, while seamlessly routing deep debugging or code generation tasks to Sonnet/Opus. This would save massive amounts of project tokens.

Additional Context

_No response_