Router
The model-routing decision as its own call: send auto, balanced or best and Conifer picks the model for each turn from the catalog the key can call, returning the pick and up to three ordered fallbacks from POST /v1/route without running a completion; the call is free and the router can be leased to coding-agent builders and companies to serve requests on their own keys.
What it does
- Model inference
Sources
- Pricing
-
Official documentation · 19 Oct 2026
"The router itself is free."; the model that serves a turn is billed at its catalog price. Leasing the router is by contact and states no price.
- Description
-
Official documentation · 19 Oct 2026
"The routing decision on its own, leased to harnesses and companies."; "POST /v1/route returns the pick and up to three fallbacks, then gets out of the way."
- What it does
-
Official documentation · 19 Oct 2026
"Send auto, balanced or best and Conifer picks the model for each turn from the catalog your key can call."
- Sold within
-
Official documentation · 19 Oct 2026
Self-serve: the route call needs only a Conifer API key minted from the console; the page states the call is free.
- Name
-
Official documentation · 19 Oct 2026
Products page entry: "04 Router".
- URL
-
Official documentation · 19 Oct 2026
Router documentation page, linked from the Products page entry "04 Router".
- Deployment
-
Official documentation · 19 Oct 2026
A hosted endpoint at api.conifer.build/v1/route called with the customer's Conifer API key.
- Interfaces
-
Official documentation · 19 Oct 2026
"POST /v1/route" with a bearer-token Conifer API key.
- Modalities
-
Official documentation · 19 Oct 2026
Input is "query" text and a "policy"; output is a JSON object with "model", "fallbacks" and "policy".
Also from Conifer
-
Conifer API service
OpenAI- and Anthropic-compatible inference gateway that puts over 200 models behind one key, picks the model and provider per request or lets the caller name one, fails over when a provider is down and returns the cost in the response; the command-line runtime and local inference are free and cloud tokens bill at the model's own rate with no upcharge.
-
Parallax Developer tool
Parallel code review for coding agents: when a session ends, models from other labs picked by Conifer's router go over what the agent wrote for bugs, security holes, dead code and slow code, each finding is proved by a test in a throwaway copy of the repository, and fixes arrive as pull requests; a Mac app (and a Linux daemon opened in a browser) from $5 a month, with unlimited DeepSeek and GLM on every plan and 5 free reviews for a new account.
-
Console Application
Observability dashboard for a team's AI use through Conifer: spend by member and by model over the last thirty days, usage, cost, routing decisions, balance and limits in one place, with a key per person, service or project that can be revoked and stops being honoured within seconds.
-
Stable Developer tool
Slash commands that live inside the coding agent already in use (Claude Code, Codex, pi): /model chooses from over 200 models or hands the choice to the router, /harness runs another harness inside the current session, and /reviewer sets any model as the reviewer for a session; early access on request, with no price published.
Something wrong here? Send a correction — quote this product id: conifer-router.