GoKawiilCerebras adds Qwen 3.8 27B model to public API at ~1500 tokens/sec
Cerebras has added the 27-billion-parameter Qwen 3.8 model to its public API endpoints, offering inference speeds of roughly 1500 tokens per second. The model supports 64k context on free tier and 128k on paid tier, and is available under Cerebras's free trial and pay-as-you-go pricing, subject to rate limits.
GoKawiil's interpretation of the reporting above, not reported fact.
Cerebras continues to differentiate itself in the inference market by pairing open-source models with unusually high throughput, a key selling point against GPU-based providers. Making a mid-sized 27B model available at these speeds gives developers a faster, cheaper option for applications needing large context windows without sacrificing accuracy, since Cerebras confirms it does not serve pruned or architecturally altered versions on public endpoints.
- Qwen 3.8 27B is now live on Cerebras's public API at about 1500 tokens/sec.
- The model offers 64k context for free-tier users and 128k for paid users.
- Cerebras states all publicly hosted models remain unpruned and architecturally unmodified, using only selective weight quantization for storage.
Source: inference-docs.cerebras.ai, 2026-09-03
Published there as: “Qwen 3.8 27B available on Cerebras at 1500 tok/SEC”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.