As recorded at collection: Free language models during the Hetzner testing period
As recorded at collection: Hetzner offers several language models, including DeepSeek-V4-Flash-0731, for free via API during the testing period. Usage limits are 3M input tokens and 60k output tokens per minute, and 500M input tokens and 5M output tokens per day.
- What you get
- Free language-model access during testing
- Price
- Free
- Valid until
- During testing
- Requirements
- Per-minute limit: 3M input tokens and 60k output tokens · Daily limit: 500M input tokens and 5M output tokens
Details
The models available for free are DeepSeek-V4-Flash-0731, GLM-5.2-NVFP4, Kimi-K2.7-Code, and Qwen/Qwen3.6-35B-A3B-FP8. They are offered during the testing period, with limits on input and output token usage.
No separate application process or coupon code was provided. You can use the models during the testing period.
You can use up to 3M input tokens and 60k output tokens per minute. The daily limit is 500M input tokens and 5M output tokens. Hetzner only retains data such as request times and token counts, which is needed to track usage and bill you in the future. It does not store request or response content, except where required by law.
Sources and verification history
- Collection channel
- linux.do
- Collected at
- August 11, 2026, 23:30 KST
- First published
- Updated