Cloudflare has graduated its Clairic extraction model from preview while simultaneously unveiling Clairic-flash, making weights for both architectures publicly available. The models originate from Cloudflare's newly formed Workers AI Superintelligence group, aimed at optimizing edge-native intelligent workloads.
Clairic is purpose-built to tackle document parsing and structured JSON extraction from complex inputs without requiring bulky foundation models. Clairic-flash provides a lightweight alternative optimized for ultra-low latency, allowing developers to execute precise parsing pipelines directly within serverless edge functions.
By releasing open weights, Cloudflare strengthens its edge computing value proposition against traditional hyperscalers. The initiative provides developers with predictable inference costs while minimizing round-trip network latency for real-time data ingestion pipelines.