Cloudflare Clairic turns messy inputs into typed values in one pass

What is changing?

Cloudflare released Clairic, its first “System One extraction” model, in preview on Workers AI on October 19. Instead of generating free-form text, the model extracts typed values such as names, dates, numbers, lists, estimates and forecasts from unstructured text and up to four images. Each scalar result includes a confidence score and the probability that the input does not determine the answer.

Clairic runs in a single forward pass and generates no output tokens. Cloudflare’s documentation lists a 65,536-token context window and pricing of $0.10 per million input tokens. The model is available through Workers bindings or the REST API, with an open-weights release planned later; the current endpoint remains an early checkpoint.

Cloudflare says one- and two-step calculations are more reliable, while long calculation chains, long-horizon forecasts and unfamiliar data need review by people or code. Clairic is therefore best treated as a fast parsing layer before an LLM in document-processing and agent workflows.