Specialist hardware
Cerebras Inference
/cerebras · Sunnyvale, USA · est. 2015
Data: mediumVisit provider ↗
Key facts
- Cerebras Inference is a specialist hardware headquartered in Sunnyvale, USA, founded 2015.
- Exposes an OpenAI-compatible API, so existing OpenAI SDK code can target it by changing the base URL.
- Blended reference price is $0.70 per 1M tokens (in $0.60 / out $0.80).
- Reference latency 110 ms TTFT at 520 tok/s throughput, with 99.5% reference uptime.
- Editorial reputation index 82/100 from 190 weighted signals; 90-day trend improving.
Specification
- Category
- Specialist hardware
- Headquarters
- Sunnyvale, USA
- Founded
- 2015
- Model access
- Open weights
- Flagship / served
- Wafer-scale served open models
- Reference open model
- Llama 3.3 70B
- OpenAI-compatible API
- Yes
- Blended $/1M
- $0.70 seed
Strengths & trade-offs
Strength
Extreme throughput from wafer-scale engines.
Trade-off
Newer public cloud; limited model and region coverage.
Profile reviewed 2026-07-10. Confidence for this listing:medium — Figure based on public sources but not re-confirmed on the exact review date; may drift.Not affiliated with Cerebras Inference. ← Back to directory