Pricing
Pay for what you use, by the minute
GPUs by the hour, open models by the token. Pay only for what you use, from prepaid credits, with no subscriptions and no hidden fees.
GPU rental
Per hour, billed by the minute. JupyterLab on the GPU you pick.
1 credit = $1 · no hidden fees
GPUVRAMCUDAFP32 TFAvailabilityPrice
Inference API
OpenAI-compatible endpoints for open models, priced per million tokens.
Per million tokens, billed per request from creditsBatch API: 50% off every price below →
| Model | Context | Input | Output | Batch |
|---|---|---|---|---|
| Chat & vision/v1/chat/completions · /v1/responses | ||||
Qwen3.8 27B qwen3.8-27b · vision | 131k | $0.12 | $1.70 | $0.06 / $0.85 |
Qwen3 4B qwen3-3b | 16k | $0.05 | $0.20 | $0.025 / $0.10 |
| Embeddings/v1/embeddings | ||||
Multilingual E5 Small (embeddings) multilingual-e5-small | 1k | $0.02 | – | $0.01 |
| Classification/v1/classify | ||||
OpenJev 4B (NLI classifier) openjev-4b | 4k | $0.045 | – | – |
| Image generation/v1/images/generations | ||||
Z-Image Turbo (1024px, 8 steps) z-image-turbo | – | $0.003 / image | – | |
Free trial: 250,000 tokens on a free key. Failed requests are never charged.Get an API key
Rendering
Blender renders on network GPUs.
Free during beta
Blender Cycles and Eevee on network GPUs. Usage-based pricing will be announced before the beta ends.
Ready to get started?
Rent a GPU, call an open model or render a scene. No commitments, pay as you go.