Claude Model Router
Claude Code mod that sets model and effort per turn and subagent from a prompt classifier: Jev by default, or Clef, OpenAI or Ollama.
Overview
Claude Model Router is a Claude Code mod that chooses a model and effort level for each main-conversation turn, advised by a prompt classifier. It changes only the model and effort in Claude Code's turn hook; Claude Code still sends requests, streams replies and runs tools, and the router does not proxy Anthropic traffic or run a local server.
The classifier is asked at the start of each logical turn for the tier (micro, low, medium or high per the listing) and the turn's activity, and again in the background every four requests. Jev is the default, asked as a third choice question in the same request; alternatives are Cloudflare Clef variants, OpenAI or a local Ollama model, and only the active classifier receives prompt text. Subagents get one model each at start when their definition does not name one, unless subagent routing is turned off to keep that text local. Configuration lives in a pane or router.json.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which tier should this turn use? | micro · low · medium · high |
| Choice | What activity is this turn? | — |
How it works
- Turn starts; classifier gets the prompt
- Choose tier and activity
- Hook sets model and effort
- Re-check every four requests; route subagents at spawn
Strengths
- Does not proxy traffic
- Swappable classifier including local Ollama
- Subagent routing can be disabled for privacy
Caveats
- Prompt text goes to the chosen classifier's service
- Requires mod support in Claude Code
Getting started
Install the mod in Claude Code and pick a classifier on the Classifier tab or in router.json.At a glance
- Type
- Proxy
- Status
- beta
- Runtime
- Claude Code mod
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Claude Code plugin that picks the model and effort for each turn.
wh000wh000/awesome-jev-liveClaude Code plugin that auto-picks the right model for each turn — micro, low, medium, or high tier — using Jev/Clef routing.
hellogumbo/awesome-jevClaude Code mod that sets the model and effort for each main-conversation turn from a prompt classifier: Jev by default, or Cloudflare Clef, OpenAI or a local Ollama model. It does not proxy Anthropic traffic.
MrJev/awesome-jev
Frequently asked questions
Does it proxy my Anthropic traffic?
No, it only changes model and effort in the turn hook.
Which classifiers are supported?
Jev (default), Cloudflare Clef variants, OpenAI and local Ollama.
Are subagents routed?
Yes, once at start, unless subagentRouting is off.
Related projects
- jev-router★ 593
Per-turn model routing for Claude Code and Codex: wraps the real CLI and lets Jev send simple turns to fast models and hard ones to strong.
ProxybetaJavaScript TypeScript/Bun launcher that asks Jev which model and reasoning effort each Codex, Claude Code or Pi turn should use.
ProjectbetaTypeScriptCLI, Claude Code plugin and Pi virtual model that pick the right model per prompt from one Jev call with a complexity floor.
PluginbetaTypeScriptClaude Code mod that picks reasoning effort per prompt with a Haiku judge, optionally Jev, and shows cache and context state.
ProjectbetaTypeScriptGo CLI model router that uses one Jev analysis to recommend a model, harness and effort level, then selects deterministically.
CLIbetaGoClaude Code plugin forked from jev-router that asks Jev how much effort each typed prompt needs and applies it to the turn.
ProjectbetaTypeScript