Conifer — run AI locally, route the rest

conifer
An ASCII luminance dither of a conifer, framed by a sun and moon. Hover with a pointer to ripple the characters along the row.

Conifer routes to the Pareto frontier

A eucalyptus sprig, drawn in ASCII.docsA single pine cone, drawn in ASCII.modelsA single leaf, drawn in ASCII.consoleA bare twig, drawn in ASCII.faqA seed head, drawn in ASCII.contactA bonsai pine, drawn in ASCII.vision

Works with the cloud, self-hosted models, and your own hardware.

  1. Install the runtime

    No charge. The runtime, the router and local inference are free.

  2. One key to the gateway

    Over 200 models behind api.conifer.build: the frontier labs and the open-weights field, Qwen, DeepSeek, GLM, Kimi, Llama and the rest. Or bring your own provider keys.

  3. No upcharge

    Cloud tokens bill at the model’s own rate and nothing on top. Your own keys and your own hardware, through any llama-compatible endpoint or our engine, cost nothing.

Dynamic query-based routing.

Works anywhere.

  • Claude Desktop
  • Codex Desktop
  • Antigravity
  • Terminal
  • cmux
  • VS Code
  • OpenCode
  • Pi
  • Zed
  • Cursor

The harness, the model, the provider and the cache, handled for you.

Each query gets those four chosen for it as it arrives. All you send is the text.

› refactor the auth middleware, keep the tests green

askanswerharnessmodelprovidercache strategy
Claude CodeKimi K3Togethercache read