# Cloudflare releases Clairic, a document extraction model for agent chains

published: 2026-10-20T12:05:49.683Z
topic: infrastructure
source: https://the-decoder.com/cloudflares-clairic-pulls-data-from-documents-faster-and-cheaper-than-llms/
url: https://aiagentnews.io/stories/cloudflare-releases-clairic-a-document-extraction-model-for-agent-chains

Cloudflare has released Clairic, a model that pulls named fields such as order numbers and invoice totals from unstructured text, and it is in preview on Workers AI. Cloudflare says it is faster and cheaper than large language models for this kind of work and is designed to sit inside a larger agent chain.

Users list the fields they want, and Clairic returns matching values in the requested format along with a confidence score. It answers all fields in a single pass rather than generating text token by token, and it returns "null" when a value is missing. It can also handle simple calculations, estimates and forecasts.

Cloudflare positions Clairic to run alongside LLMs and other "System One" models. In its example, a complaint email goes through Clairic to extract the order number, amount and damage and estimate the repair cost, then the Clef decision model applies company rules to choose a refund or repair, and GLM-5.3 writes the customer reply. Cloudflare says that chain takes 19.3 seconds, while GLM-5.3 alone needs 61.7 seconds for the three steps and costs almost twice as much.

Cloudflare says Clairic cannot yet reliably handle longer chains of calculations, so developers should run those in their own code for now. Open weights are set to follow soon.

## Why it matters if you build agents

Developers building extraction steps into agents can use a single-pass model for fields like order numbers and totals, with confidence scores to help check results. Multi-step arithmetic should stay in application code for now.

The speed and cost comparisons are Cloudflare's own, so test on your own documents before relying on them.