Log in to your Hetzner account, or register a free one.
Open Apps in the left sidebar, then select Inference.
Click Create API Token in the top right.
Copy the generated token right away. It is shown only once, so keep it somewhere safe.
Wrapping Up
Use the base URL https://inference.hetzner.com/api/v1 with the model Qwen/Qwen3.6-35B-A3B-FP8. It is 100% OpenAI-compatible, so it works as a custom provider in OpenCode, Cursor, or your own scripts. Expect slower latency on the shared free tier, but it is hard to beat for lightweight tasks.
Comments
Join the conversation — sign in to comment.
No comments yet — start the conversation!