Cloudflare has announced the release of Clef-omni, a new multimodal decision model built on the Qwen3-Omni-30B-A3B architecture and distributed under the Apache-2.0 license. According to reporting by MarkTechPost, the model is capable of processing and scoring text, images, audio, and video inputs within a single API call. Pricing is set at $0.15 per million input tokens for Clef-omni, alongside a lighter variant named Clef-flash priced at $0.038 per million input tokens.
Multimodal Processing and Open-Source Licensing
The release marks an expansion of Cloudflare’s AI offerings into native multimodal decision-making, handling diverse data types simultaneously. By utilizing an Apache-2.0 license, the company provides developers with open access to modify and deploy the architecture within their own infrastructure. The tiered pricing structure between Clef-omni and Clef-flash suggests an effort to accommodate both high-accuracy decision tasks and lower-cost, high-volume processing demands.
Unknown Technical Specifications
Despite the initial announcement, several key technical details remain undisclosed. The available information does not specify the exact hardware requirements needed to self-host Clef-omni, nor does it provide benchmark scores comparing its decision-making capabilities against other models in the Qwen or open-source ecosystem. Furthermore, the announcement did not specify output token pricing, latency metrics, or the exact architectural modifications made to the underlying Qwen3-Omni base. Developers will need to monitor future documentation for performance evaluations and comprehensive API integration guides.
As these models deploy to production environments, further testing will clarify how the pricing and performance translate to real-world multimedia evaluation tasks.
Lead image: panumas nikhomkhai / Pexels
