Model updates

Official news, first rumor reports and substantive progress. Each entry keeps the information published at that time.

• Cloudflare launches Clef-omni multimodal decision model, cuts Clef-flash price, and speeds hosted Clef

  Cloudflare’s October 9 blog introduces Clef-omni, an Apache-2.0 30B-A3B MoE decision model post-trained from Qwen3-Omni that scores text, images, audio, and video in one schema-bound call on Workers AI at $0.15 per million input tokens (64k context), with open weights on Hugging Face. The same update drops hosted Clef-flash to $0.038 per million input tokens with a 24k hosted context tradeoff, and reports roughly 1.7×–2.0× faster median latency for hosted Clef after moving serving to SGLang.

  OfficialOfficial updateOct 11Source: Cloudflare BlogPart of Clef
• Cloudflare's Clef decision model now runs via llama.cpp with official GGUF weights

  Clef, Cloudflare's decision model, can now run via llama.cpp using official GGUF weights published under ggml-org/Clef-GGUF, with a one-command llama serve launch. A community speed test compared it against Ollama on an M5 Max MacBook Pro. This follows Clef's earlier arrival on Ollama, extending the decision model stack to llama.cpp users.

  OfficialOfficial updateOct 6Source: X @ClementDelanguePart of Clef

RSS for this view →