jev-model-router (zerone-agents)
Go model-routing gateway for OpenAI and Anthropic clients where Jev picks a model from editable model cards, with CLI and dashboard.
Overview
Jev Model Router from zerone-agents is a lightweight gateway that agents connect to with OpenAI Chat Completions or Anthropic Messages clients. A request can name an explicit model, or Jev can choose one from editable model cards using a routing prompt. Providers are managed through a CLI and an English/Chinese dashboard.
It runs as a single Go service with SQLite, streams text and tool calls, and supports OpenAI-compatible and native Anthropic upstreams within a documented compatibility matrix; unsupported conversions fail explicitly. The dashboard shows instance status, models, the routing prompt and routing records, and lets you edit model descriptions. Separate settings and inference tokens are required at startup. The recommended deployment is a prebuilt Docker Compose stack. Local Laya routing, sensitive-session locking and PostgreSQL support are listed as planned. The current release noted is v0.1.19.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which model card best fits this request? | — |
How it works
- Clients send OpenAI or Anthropic-style requests to the gateway
- If no model is named, Jev chooses from the model cards using the routing prompt
- The request is forwarded to the chosen upstream and streamed back
- Routing records are stored and visible in the dashboard
Strengths
- Drop-in for OpenAI and Anthropic client formats
- Single Go binary with SQLite, Docker Compose quickstart
- Editable model cards and routing prompt without code changes
- Explicit failures for unsupported conversions
Caveats
- Compatibility limited to the documented matrix
- Anthropic /v1/messages to native Anthropic upstream is still planned
- Early release series (v0.1.x)
Getting started
Deploy with the Docker Compose quickstart, or go build -o /tmp/jev-router ./cmd/jev-router and run /tmp/jev-router serve with the two required tokens.At a glance
- Type
- CLI
- Status
- beta
- Language
- Go
- Runtime
- Gateway service with CLI and web dashboard
- Has code
- yes
Use cases and integrations
How curated lists describe it
Lightweight model routing for agents, with OpenAI and Anthropic API…
wh000wh000/awesome-jev-liveLightweight model routing for agents, with OpenAI and Anthropic API compatibility, automatic model selection, and a management CLI.
hellogumbo/awesome-jev
Frequently asked questions
What does Jev decide?
Which configured model card should serve a request when the client does not name a model.
Which client formats are supported?
OpenAI Chat Completions and Anthropic Messages, within the compatibility matrix.
Can it run local decision models?
Local Laya routing is listed as planned, not yet available.
Related projects
OpenAI-compatible LLM router across 80+ providers with an optional jev-auto mode that picks the model using calibrated probabilities.
LibrarybetaTypeScriptGateway that distils System One or LLM answers into small fine-tuned classifiers and serves them when confident, falling back otherwise.
ServiceprototypeRustLocal-first routing engine for long-running agent tasks that uses a Jev decider to pick a model group per turn.
PluginproductionRust, TypeScriptOpen-source AI gateway for 100+ LLMs that can use Jev to classify requests for complexity routing and to compact tool results.
ProxyproductionPythonOpen-source workspace for running and reviewing AI coding agents across devices, with an optional Jev-based automatic model router.
ProjectproductionTypeScript- 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