16 October 2026

Cloudflare adds audio, video and image input to its decision models

First reported

Deep Learning Weekly ran this on 16 October.

Cloudflare's Clef decision models can now take audio, video and images as input, not just text.
The new Clef-omni version is built on Qwen3-Omni-30B-A3B, a multimodal model that handles several kinds of input at once.
Text decisions come back in about 130 milliseconds at the median, a fast response time.

Our read

Only Deep Learning Weekly picked this up, the sole newsletter of the 30 to cover it. It is the Cloudflare story that adds audio, video and images to the input side, which the archive's earlier multimodal items did not address.

Earlier on this story

Aug 26EchoWM model generates video, audio, and speech from camera movements
Aug 21Adobe adds three AI audio tools to Firefly creative platform

Written by Newskeryx from which newsletters covered this story and from our own archive.

How it was covered

Deep Learning WeeklyEditorial team

Clef-omni multimodal launch, Clef and Clef-flash variants, base model, and latency