Cloudflare open-sources Clairic document-extraction model weights
11/21/2026 — 11/21, 22:17·1 sources (all)·1 reports (all)
Key takeaways
- Cloudflare released the weights of Clairic and Clairic-flash on Hugging Face under Apache-2.0, at 27B and 9B parameters respectively and bui…
Background & analysis
On 18 November 2026, Cloudflare published the open weights for two document extraction models, Clairic and Clairic-flash, on Hugging Face under the Apache-2.0 licence. Clairic is the 27B-parameter variant, while Clairic-flash is a 9B-parameter release built for lower-latency workloads. According to the model cards, Clairic is based on Qwen3.8-27B and Clairic-flash on Qwen3.5-9B.
Both models address the same task. They take a schema supplied by the caller, read unstructured text, images and PDF pages, and convert that input into structured, typed fields, returning typed values, estimates and forecasts that a program can consume. Opening the weights did not end Cloudflare's hosted offering, which remains available on Workers AI through the endpoints @cf/cloudflare/clairic and @cf/cloudflare/clairic-flash for callers who prefer managed access.
Public coverage of the release followed on 21 November 2026. Cloudflare positions the move as a step toward letting a company act as a neolab, building its own frontier models and serving them to its own customers, and frames the two Clairic releases as a focused instance of that ambition, since they target a single job, turning documents into fields a program can read, and compete mainly on cost and latency. That is where the story stands at present: the weights are out under Apache-2.0 and the hosted endpoints on Workers AI are still live.
AI-generated from 1 reports · updated 12 hours ago
- Heat index
- 70
- All sources
- 1
- All reports
- 1
- First seen
- 13 hours ago
Reports on this story headlines open the original
Cloudflare has released open weights for Clairic and Clairic-flash, its 27B and 9B document-extraction models, on Hugging Face under Apache-2.0, with hosted endpoints staying on Workers AI. Both take a caller-supplied schema and turn unstructured text, images, and PDF pages into typed fields; the model cards name Qwen3.8-27B and Qwen3.5-9B as base models.
DEV Community · AIAI score 56
Other stories people are talking about
- Heat index 62Cloudflare open-sources Kitesurf, a browser engine built for AI agents1 sources
- Heat index 1143RisingAnthropic Reaches $1.6 Trillion Market Value in First Trading Week After IPO21 sources
- Heat index 1011Anthropic Mathematician Says Claude Formalized the Generalized Riemann Hypothesis21 sources
- Heat index 622Anthropic IPO total raise reaches $93.5 billion after greenshoe exercise20 sources
- Heat index 596RisingThinking Machines Lab Open-Sources Inkling 2 Mini with 32GB Model Card Recommendation10 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion

