TypeSafe: Jev Router

Jev Router picks the best model and reasoning effort for each request, balancing quality, speed, and cost. It runs on Jev, TypeSafe's first System One model, and adapts as your conversation evolves. One endpoint gives you the whole model ecosystem, leveraging the ecosystem's token and spend share.

The Jev Router documentation on OpenRouter includes the Jev Tutorial guide, the Jev (TypeSafe) guide, the TypeSafe SDK (Jev) guide, the Jev Router guide, the Gate Agent Tool Calls with Jev cookbook, the Auto-Approve Coding Agent Permission Prompts with Jev cookbook, the Classify Reddit and YouTube Comments with Jev and ScrapeCreators cookbook, the Classify and Tag Text at Scale with Jev cookbook, the Cut LLM Cost with a Jev-Verified Cascade cookbook, and the interactive Jev Lab. The quickstart covers API keys and your first request, and the API reference lists every parameter.

Jev Router has no fixed price of its own. Each request is billed at the price of the model that serves it, so your cost depends on which model the request is routed to.

Jev Router can route to models with context windows of up to 1,000,000 tokens. The context length available to a request is that of the model that serves it.

Jev Router accepts audio, files such as PDFs, images, text, and video as input and returns text.

Jev 1.14 and Jev 1.13 are other text models from TypeSafe.

Jev Router was released on September 25, 2026.