Article
Cloudflare enables 3 AI to handle claims in relay: self-testing is 3.2 times faster, with costs reduced by 49%Beating News:AI Cloudflare has released a preview of Clairic, a data extraction model. Give it an invoice or an after-sales email, and it can identify the order number, date, and total amount; it can also estimate repair costs and forecast customer traffic based on available information.
Unlike conventional LLM, it does not generate answers Token one at a time; instead, it directly computes and returns results of the specified type, along with confidence levels. Fields for which no supporting evidence can be found can return null.
Clairic follows the System One model approach that gained traction after TypeSafe launched Jev. Cloudflare has already launched its decision model Clef on October 1, which selects from preset options; Clairic extracts, calculates, and forecasts specific numerical values. Neither needs to generate answers word by word. Clairic charges only Token per million input 0.10 USD, with no output Token fee.
Cloudflare demonstrated product-damage claims processing: Clairic first extracts information and estimates repair costs, Clef decides between a refund and repair, and GLM-5.3 then replies to the customer. Together, the take 19.3 seconds, while assigning everything to GLM-5.3 takes 61.7 seconds. On a per-million-claim basis, costs fall from 11740 USD to 6032 USD, a reduction of 49%. For information extraction alone, processing time falls from 16.1 seconds to 0.46 seconds.
However, although Clairic does not generate output Token, when fields cannot be handled in a single computation, the system automatically splits them into multiple parallel rounds, with each round rereading the full input. For high Token inputs such as images, this may increase actual invocation costs.
The model is currently available in preview on Workers AI, and its weights have not yet been open-sourced.
Original link https://m.theblockbeats.info/flash/372790