ZeroGPU
Run specialized and open-weight models faster and at lower cost across ZeroGPU edge and cloud infrastructure.
ZeroGPU sells inference on Liquid Inference. A buyer names one of its models as zerogpu/<model>, or names the model and pays the lowest offer that meets the buyer's rules.
Facts
- Handle
- zerogpu
- Website
- https://zerogpu.ai/
- Terms of service
- https://zerogpu.ai/terms
- Privacy policy
- https://zerogpu.ai/privacy-policy
- Status page
- https://status.zerogpu.ai/
- On the exchange since
- 2026-11-03
- Models
- 3
- Offers
- 3
- Regions
- us
- Precisions (declared)
- bf16, fp8
- Lowest price on
- 0 of 3 models, for the agent mix
Reliability, last 3 days
- Uptime
- 94.9%
- Hours with an answer
- 100.0%
- Median response time
- 2.5s
Checked by the exchange
0 of 3 offers match the fingerprint of the model. 0 have a finished evaluation.
The exchange sends fixed prompts to each offer and compares the answers with a baseline for the model. It also runs benchmarks against the offers and keeps the results.
Data handling
- Zero retention, declared by the provider
- No training, declared by the provider
Models
- DeepSeek V4.1 Flash 1017: us, declared precision fp8, $0.22 input and $0.66 output per million tokens, $0.0066 cache read per million tokens, 1M tokens of context
- GLM 5.3 Flash: us, declared precision fp8, $0.099 input and $0.33 output per million tokens, $0.0198 cache read per million tokens, 1M tokens of context
- GPT: OSS 120B: us, declared precision bf16, $0.0407 input and $0.187 output per million tokens, 131.1K tokens of context