Which models are supported?
v1 supports all models served through the OpenAI Chat Completions protocol on OpenCode Zen: DeepSeek, MiniMax, GLM, Kimi, Big Pickle, and all *-free models. GPT, Grok, Claude, Qwen, and Gemini use different protocols and are planned for a future release.
Why is my model missing from the dropdown?
The dropdown is filtered to OpenAI-compatible models only. If a model uses the Responses, Anthropic, or Google protocol it will be hidden — this is intentional in v1.
Is this affiliated with OpenCode / Anomaly?
No. This is a community project by pranjal-joshi. It uses the public OpenCode Zen API and is not officially affiliated with OpenCode or Anomaly.
How are tokens billed?
Billing is handled entirely by OpenCode Zen through your account at opencode.ai/auth. This integration does not charge or collect any money.
What is the free model usage policy?
Free models may use your data to improve the models and may have trial limits. Do not submit personal or confidential data to free models. See the OpenCode Zen docs for details.
Can I use my own endpoint?
Yes. During setup you can override the Endpoint field (defaults to https://opencode.ai/zen/v1), which lets you point at a custom gateway.
Is there web search support?
Not in v1. OpenCode Zen doesn't expose a web-search option like OpenRouter does. If the model supports it natively, the prompt can suggest it.
How do I report a bug?
Open an issue on the GitHub issue tracker with your HA version, HACS version, model, and relevant logs. Never paste your API key.