The new System‑One model, Clairic, can pull structured information from documents in one go, delivering faster and more cost‑effective extraction than conventional large language models.

Cloudflare has unveiled Clairic, its latest System‑One model that extracts structured data from documents in a single pass, promising faster and more cost‑effective results than traditional large language models.

What is Clairic?

Clairic is a token‑free model designed to pull out tables, forms, and key‑value pairs from PDFs, scans, and other document formats without the need for multiple inference steps.

How it differs from conventional LLMs

Typical large language models process text token by token, which can be slow and expensive for lengthy documents. Clairic bypasses this by analyzing the entire document in one go, reducing compute overhead and cutting costs.

The model also sidesteps the token limit constraints that often force developers to chunk documents, a process that can introduce errors and inconsistencies.

Potential use cases

  • Automated invoice processing
  • Legal contract analysis
  • Healthcare record extraction
  • Financial statement summarization

Enterprises can integrate Clairic via Cloudflare’s API, enabling real‑time data extraction for workflows that previously required manual review or expensive custom pipelines.

While the model is currently offered as part of Cloudflare’s broader AI suite, the company says it will expand access and add fine‑tuning capabilities based on customer feedback.

For more details on Cloudflare’s product roadmap, see the Product Design section on uxc.news.

The Decoder coverage of Cloudflare’s Clairic launch

Track coverage on Momentum · More in Product & Design.