Skip to content
NewA calmer, faster ZeroHour
StoryModel releaseinfo

Cloudflare previews Clairic extraction model on Workers AI

4 sources · 4 articlesFirst reported Updated ()1 · 1 read
What's new

October 20: Help Net Security and The Decoder add claims-workflow cost, latency, and reliability limits beyond the October 19 preview announcement.

Merged summary · x-ai/grok-4.7 · rewritten as coverage arrivesSummaries by Vani AI

Cloudflare’s Clairic extraction model is in Workers AI preview, with open weights planned and no output-token charge.

Cloudflare announced a preview of Clairic, which it calls the first System One extraction model, on Workers AI as @cf/cloudflare/clairic. It returns typed spans, numbers, dates, lists, or distributions with confidence scores, or null, from text and up to four images in one pass without generating tokens. Cloudflare says responses typically take under a second, charges only for input, and intends to release the weights later. In a damaged-product claims test, Help Net Security reports a Clairic cascade at 19.3 seconds and $6,032 per million claims versus 61.7 seconds and $11,740 for an all-LLM version. The Decoder frames the same times as a Clef and GLM-5.3 chain versus GLM-5.3 alone. Cloudflare calls the endpoint an early checkpoint and says longer arithmetic should stay in user code.

Key facts
  • Model id is @cf/cloudflare/clairic; open weights are planned, not released.
  • Input-only billing; Cloudflare says typical responses take under a second.
  • Claims cascade: 19.3 seconds and $6,032 per million claims, about 49% cheaper.
  • Clairic claim extraction took 0.46 seconds; Clef decision took 0.22 seconds.
  • Airport-count forecasts missed by 3.3% on average over 21 days.
  • Decision models Jev and Clef remain separate from this extraction class.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

Timeline

dated events in the reporting · oldest first
  1. Oct 19, 2026Cloudflare announces Clairic preview on Workers AI · Cloudflare Blog
  2. Oct 20, 2026Help Net Security publishes claims-workflow cost and latency figures · Help Net Security
  3. Oct 20, 2026The Decoder reports sample-chain timing versus GLM-5.3 · The Decoder

What each outlet adds

beyond the shared facts
  • Cloudflare BlogModel id, input-only billing, output types, and Jev/Clef split
  • Help Net SecurityPer-million-claims costs, step timings, and 3.3% forecast miss
  • The DecoderFrames 19.3 versus 61.7 seconds as GLM-5.3 comparison
  • MarkTechPostFetched body is only a cookie notice, no model details
Vendors
Cloudflare
Products
Workers AI
Organizations
TypeSafe
AI models
ClairicGLM-5.3ClefJevGemma 4Qwen

Coverage

oldest first · one row per article · UTC
  1. · 6d ago Cloudflare BlogFirst report

    Cloudflare previews Clairic, a System One extraction model

    Clairic is hosted on Workers AI as @cf/cloudflare/clairic, charges only for input, and open weights are planned.

  2. · 5d ago Help Net Security

    Cloudflare previews Clairic extraction model that skips token generation

    In one claims workflow, a Clairic pipeline finished in 19.3 seconds and cost 49% less than an all-LLM version.

  3. · 5d ago MarkTechPost

    Cloudflare releases Clairic extraction model on Workers AI

    The captured page is only a cookie notice, so model size, benchmarks and access are not stated.

  4. · 5d ago The Decoder

    Cloudflare previews Clairic, a single-pass document extraction model

    Clairic is in preview on Workers AI, with open weights planned, and returns named fields plus a confidence score.