Clairic
Cloudflare's first System One extraction model returns typed values, estimates and forecasts in one pass with no output tokens.
Overview
Clairic, introduced by Cloudflare on October 19, 2026, is its first System One extraction model. It returns typed values from unstructured input in a single pass without generating reasoning or output tokens: extracted text spans with offsets and confidence, computed numbers, dates and durations, estimates with a median and 80% interval, and per-step forecasts. It accepts text plus up to four images and returns null with p_null when no answer exists. It is a preview on Workers AI as @cf/cloudflare/clairic, billed on input only, and Cloudflare says open weights will follow soon. In Cloudflare's own claims-pipeline example, Clairic plus Clef plus GLM-5.3 took 19.3 s versus 61.7 s for GLM-5.3 alone, at about 49% lower cost.
This is a sourced directory entry. Features and capabilities are reported in the linked sources; this entry does not document independent Aymorphic testing.
Key features
- Single-pass extraction, computation, estimation and forecasting without output tokens
- Estimates and forecasts return p10-p90 intervals
- Text plus up to four optional images; null with p_null when unanswerable
- Preview on Workers AI via env.AI.run() or REST
- Cloudflare example forecast: 3.3% average error over 21 days, 17 of 21 actuals inside the 80% interval
Use cases
- Pulling fields from receipts, claims and forms before an LLM step
- Quick estimates and basic forecasts where a dedicated time-series model is overkill
Pricing
Preview; Cloudflare charges only for input tokens with no output-token charge. A per-token rate is not stated.
Pricing reflects the source record and may change. Confirm current terms with the provider.
Dates and sources
- Released
- 19 Oct 2026
- Record updated
- 27 Oct 2026
- Record last checked
- 27 Oct 2026