---
title: "Introducing Clairic, a new System One class of extraction models"
description: "Cloudflare announces Clairic, a preview extraction model hosted on Workers AI, designed as a System One model for fast, one-pass value extraction from unstructured inputs. Unlike decision models such as Jev or Clef that output probabilities, Clairic returns extracted, computed, estimated, or forecasted typed values like numbers, dates, durations, and spans with character offsets and confidence scores. It processes only input tokens in a single forward pass, resulting in under-one-second responses and no output token charges, ensuring every answer is a valid typed value or null. The model handles extraction tasks like invoice numbers, estimation with distributions providing a median and 80% interval, and forecasting per future step with intervals, all without token generation. Cloudflare plans to release Clairic as open-weights soon and offers it via Workers AI with a typed field request structure accessible through env.AI.run() or the REST API."
---

# Introducing Clairic, a new System One class of extraction models

[Cloudflare](https://yomu.fyi/company/cloudflare) · Kevin Flansburg · Oct 19, 2026

**Type:** Announcement

## Summary

Cloudflare announces Clairic, a preview extraction model hosted on Workers AI, designed as a System One model for fast, one-pass value extraction from unstructured inputs. Unlike decision models such as Jev or Clef that output probabilities, Clairic returns extracted, computed, estimated, or forecasted typed values like numbers, dates, durations, and spans with character offsets and confidence scores. It processes only input tokens in a single forward pass, resulting in under-one-second responses and no output token charges, ensuring every answer is a valid typed value or null. The model handles extraction tasks like invoice numbers, estimation with distributions providing a median and 80% interval, and forecasting per future step with intervals, all without token generation. Cloudflare plans to release Clairic as open-weights soon and offers it via Workers AI with a typed field request structure accessible through env.AI.run() or the REST API.

## Context

System One decision models from TypeSafe (e.g., Jev, Clef) showed purpose-fit models outperform LLMs on choice-based tasks, but extraction tasks like invoice totals or delivery dates were still handled by costly, slow LLMs. Cloudflare applied the System One design to extraction, creating Clairic.

## Approach / What changed

Clairic is trained as a System One extraction model that takes unstructured text and a typed field schema, processes input in a single forward pass without output tokens, and returns extracted spans, computed numbers, estimated distributions with median and 80% interval, and forecasted values per step, each with calibrated confidence scores.

## Takeaways

- Clairic extracts, computes, estimates, and forecasts typed values (numbers, dates, durations, spans) from unstructured inputs in a single forward pass, returning confidence scores and character offsets for each result.
- Unlike LLMs that charge for output tokens and may produce invalid responses, Clairic generates no output tokens, ensuring every answer is a valid typed value or null, with responses typically under one second.
- Clairic is designed for extraction tasks where LLMs are inefficient, such as invoice data extraction, delivery date estimation, and traffic forecasting, and is available as a preview on Workers AI with plans for open-weight release.

## FAQ

### How does Clairic differ from an LLM for extraction tasks?

Clairic processes input in a single forward pass without generating output tokens, returning typed values with confidence scores instead of freeform text. This makes it faster (under one second) and cheaper, while ensuring answers are always valid typed values or null, reducing post-processing needs.

### What types of values can Clairic return?

Clairic returns text spans with character offsets, numbers, dates, times, datetimes, durations, lists, estimates with distributions (median and 80% interval), and forecasts per step with intervals. Each scalar answer includes p\_null and confidence scores.

### How do I use Clairic from Workers AI?

You can call Clairic using env.AI.run() in a Worker or via the REST API. A request consists of a state (text and optional images) and a map of typed fields with names, types, and optional instructions.

**Tags:** [AI](https://yomu.fyi/topic/ai), [Clairic](https://yomu.fyi/topic/clairic), [Extraction Models](https://yomu.fyi/topic/extraction-models), [System One Models](https://yomu.fyi/topic/system-one-models), [Workers AI](https://yomu.fyi/topic/workers-ai)

- Source: [Cloudflare](https://blog.cloudflare.com/clairic-extraction-model)
- Source URL: https://blog.cloudflare.com/clairic-extraction-model
- Ingested by Yomu: 2026-10-19T18:00:13.861Z

[Read original post](https://blog.cloudflare.com/clairic-extraction-model)