How to Use Ox Alpha Free Right Now
Ox Alpha is currently listed at zero input and output cost, but “free” means something different on each surface. Here are the access paths we could verify today.
1. OpenRouter — direct model access
Use model ID stealth/ox-alpha in OpenRouter's chat UI or OpenAI-compatible API. The official model record currently reports $0 prompt and completion pricing, 1,048,576 context tokens, and 131,072 maximum completion tokens.
2. OpenCode — verify the live free alias
Install OpenCode, connect a provider, and run /models. OpenCode's live Zen model endpoint currently exposes x-preview-f-free. Its public endpoint does not label that ID “Ox Alpha,” so treat the mapping as unconfirmed and verify the live picker before using it.
3. Command Code — free model, paid harness
Command Code's official model page lists stealth/ox-alpha at $0 for input, output, and cache reads while the stealth preview lasts. The same page says the model is available on Go and higher, so the model request is free but the Command Code harness is not necessarily free.
Do not confuse free tokens with free risk
Use synthetic or public code first. Confirm the active provider's data policy, rate limits, and retention terms before sending secrets, customer data, or proprietary source code.
A reliable test sequence
- Run the same small task through OpenRouter first so you have a stable model ID.
- Try the OpenCode alias only after confirming it appears in the current picker.
- Record interruptions, retries, latency, tool-call errors, and total tokens—not just whether the final answer looked good.
- Keep a named fallback model configured because stealth previews can be renamed, rate-limited, or removed.