Clairic
Released Oct 19, 2026 · knowledge cutoff unpublished
- Status
- Unknown
- Location
- United States
- Type
- System One
- Modality
- Multimodal
- Context window
- 66K
- Max output
- —
- Speed
- —
- Price ($/MTok in / out)
- $0.10 / $0.00
- Cost per task
- —
- Open weights
- No
Strengths & weaknesses
Strengths
- Returns a value of the requested type or an explicit null for every field in one pass — nothing to parse, and a value the input doesn't contain comes back as absent rather than invented
- Attaches a null probability and a confidence to each scalar, and a median with an 80% interval to estimates and forecasts, giving an automation threshold to work against
- Reads up to four images alongside the text, so a receipt photo or screenshot can be the source document
Weaknesses
- Arithmetic is only dependable for one or two steps — longer computed fields need checking downstream
- Forecasts flatten toward the average a few steps out and cannot anticipate events the series doesn't show, such as holidays — not a replacement for a time-series model trained on your own data
- A preview with open weights only promised and no base model or licence named — and Cloudflare itself stresses every answer is well formed, not necessarily correct
Not an LLM: a "System One" extraction model, the first of a new class alongside Cloudflare's Clef decision models, that reads a text state plus up to four images and returns typed values (text, number, date/time, duration, list, bounded estimate, forecast up to 64 steps) in a single pass without generating tokens, so none of this site's benchmarks apply and every cell is null by design. Released 2026-10-19 as a preview on Workers AI at $0.10 per million input tokens (9,091 neurons); Cloudflare states there is no output-token charge because it generates no tokens, hence 0 as on Jev. The context window is 65,536 tokens, with fields that don't fit beside the state in one 8,192-token pass split across parallel passes. Cloudflare's launch post compares latency and cost against a GLM-5.3 pipeline and reports 17 of 21 forecast actuals inside the 80% interval, both its own figures. Open weights are planned but not yet released; base model, parameter count and knowledge cutoff are unpublished. No Artificial Analysis coverage.
Benchmarks0 of 10 reported
MMLU-Proretired
GPQA Diamond
SWE-bench Verified
SWE-bench Pro
Terminal-Bench 2.1
AIME (latest)retired
Humanity's Last Exam
LMArena Elo
GDPval-AA v2
ARC-AGI-2
For developers
API model strings
- cloudflare
@cf/cloudflare/clairic
Licence
Proprietary — weights not released
Retirement
No retirement announced
Lineage
No recorded predecessor