Instant reasoning is here on
Qwen3 32B
Up to 2,400 t/s, only on Cerebras Inference. With hybrid reasoning modes, agentic support, and advanced tool calling Qwen3-32B, by Alibaba, outperforms GPT-4.1 and Claude Sonnet 3.7— but runs faster, open-weight, and ready to deploy.
Try it today
