Cloudflare Integrates Clef AI Into SGLang for Sub-40 ms Latency

Cloudflare has integrated its Clef family of decision models into the SGLang framework, enabling sub-40 ms decision latency for the Clef-Flash model on a single H200 GPU. Developers can now access these models via the /v1/systemone API. Cloudflare also lowered the price of the Clef-Flash model and is preparing to add support for the multimodal Clef-Omni model.