Conifer

Conifer is a model gateway and local runtime for AI coding that routes each request to the cheapest model able to handle it. A free runtime installs on Mac, Windows or Linux, and a single API key reaches more than 200 models at api.conifer.build, covering the frontier labs and open weights families such as Qwen, DeepSeek, GLM, Kimi and Llama. For each query the router chooses the harness, model, provider and cache strategy, so the user only sends the text. It is aimed at developers who work across several coding agents and model providers and want to control cost without giving up quality. Conifer works with Claude Desktop, Codex Desktop, VS Code, Cursor, Zed, OpenCode and the terminal, and it can run models on your own hardware through any llama compatible endpoint. Two products sit on top of the gateway. Stable, in early access, adds slash commands inside an existing coding agent to switch models, run another harness inside the session or assign a reviewer model. Parallax runs a parallel code review by models from other labs once a session ends, and reports a finding only when a test fails on your code, passes with the fix and leaves the rest of the suite green. The runtime, router and local inference are free. Cloud usage is prepaid pay as you go, billed per token at each model's own published rate with no markup, and requests on your own OpenAI, Anthropic or Azure keys carry no Conifer fee. Parallax includes 10 free review episodes per account; later episodes use Parallax credits, which are priced separately from gateway credits. The models page publishes each model's blended price per million tokens alongside a measured intelligence score.