Heabsy
EU inference API on its own GPUs in the EEA, with zero data retention and OpenAI- and Anthropic-compatible endpoints.
Heabsy is a Slovak inference API (FEYA, s.r.o.) that serves Qwen3.8 27B on its own GPUs inside the EEA, with prompts and completions never logged. It is OpenAI- and Anthropic-compatible, so Claude Code and Cursor can point at it, and reports fresh, cached and output tokens per request.
The flagship model costs $0.04 per 1M input tokens and $0.30 output. A routed catalog of other open models is billed on the same EU invoice, labelled as routed since that compute may run outside the EEA. Accounts are opened on request with per-key budgets.
Pricing: Per token usage
Heabsy prices by model
Per 1M tokens, read off Heabsy's own pricing page on the date shown.
| Model | Input / 1M | Cached / 1M | Output / 1M | Checked | Notes |
|---|---|---|---|---|---|
| GLM 5.3 Flash | $0.04 | $0.01 | $0.12 | 29 Sep 2026 | Routed through a third-party provider; compute may run outside the EEA |
| Qwen3.8 27B | $0.04 | $0.04 | $0.30 | 29 Sep 2026 | Own GPUs in the EEA; listed as Qwen3.8-27B-NVFP4 |
| gpt-oss-20b | $0.05 | – | $0.20 | 29 Sep 2026 | Routed through a third-party provider; compute may run outside the EEA |
Heabsy Alternatives
Explore 126 products in the Inference APIs category. View all Heabsy alternatives.
DeepInfra
Run the top AI models using a simple API, pay per use. Low cost, scalable and production ready infrastructure.
CheapestInference
Flat-rate unlimited inference on open-weight models, sold in daily 8-hour windows
Clef
Cloudflare's open decision models: typed choices, scores and yes/no answers with probabilities, hosted or self-run.
Work on Heabsy? Feature it at the top of Inference APIs.
Is your product missing?