Clef-omni
Cloudflare · Released 9 Oct 2026
Multimodal addition to the Clef family that accepts text, images, audio, and video in one request and returns schema-bound, calibrated decisions. Built on a Qwen3-Omni-30B-A3B-Instruct mixture-of-experts backbone with the text-to-speech output components removed.
- Maker
- Cloudflare
- Release date
- 9 Oct 2026
- First seen
- 9 Oct 2026
- Access
- Workers AI endpoint @cf/cloudflare/clef-omni; open weights on Hugging Face.
- Pricing (as publicly stated)
- As stated in the 9 Oct 2026 announcement: $0.15 per million input tokens. See Cloudflare's developer docs for current rates, including how media is converted to input tokens.
- Modalities
- text, images, audio, video
- Question types
- noul, choice, score
- Latency (as stated)
- Cloudflare said text-only decisions return in about 130 ms at the median, image inputs in about 150 ms, audio clips in a few hundred milliseconds, and a full 21-second video clip with sound in about 1.5 seconds.
- Official announcement
- https://blog.cloudflare.com/clef-faster-cheaper-multimodal/
Benchmark claims
These are claims as published by the maker or in the linked official post. We have not independently rerun every benchmark.
Cloudflare published benchmark tables for Clef-omni against Clef, Clef-flash, and Jev, including BFCL, ToolRet, API-Bank, and several workflow evals. For example, it listed Clef-omni at 98.2 on BFCL case exact and 94.8 macro-F1 on BANKING77.
Names and marks belong to their owners. This page is an independent summary, not a product page from Cloudflare.