mlx-optiq

Use with coding agents

One command runs Claude Code, Codex, OpenCode, Mistral Vibe, Hermes Agent or OpenClaw on an OptiQ Cloud model, through optiq serve on your Mac, which holds the key.

$ optiq run claude --cloud glm-5.3-flash
$ optiq run codex --cloud deepseek-v4.1-flash

The same works for opencode, vibe, hermes and openclaw. optiq run starts optiq serve --cloud-only on your Mac, which loads no local model, or reuses an optiq serve that is already running. It configures the agent for that session only and leaves the agent's own config files alone.

The agent talks to optiq serve, which holds the key, so the agent never sees it. Any client pointed at optiq serve can also pick a cloud model by name once you are signed in:

ClientModel to pick
Claude Code, OpenClawoptiq-cloud/glm-5.3-flash on /v1/messages
Codexoptiq-cloud/glm-5.3-flash on /v1/responses
OpenCode, Mistral Vibe, Hermes Agent, anything OpenAI-compatibleoptiq-cloud/deepseek-v4.1-flash on /v1/chat/completions
OptiQ Labchoose the model under OptiQ Cloud in the chat's model list

Only clients on the same Mac can use cloud models through optiq serve. It checks the shape of a token, not the token, so on --host 0.0.0.0 anyone on your network could otherwise spend your balance. See Integrations for each client's setup.