Key Info
Clef's Omni decision model is now live on OpenRouter with native audio, video, image, and text input in a single call, returning probabilities instead of generated tokens. Pricing is $0.15 per million input tokens with output free.
Highlights
- Four modalities (audio, video, image, text) processed in one request, using the same state format as Chat Completions (image_url, input_audio, video_url)
- Returns probabilities rather than tokens; a demo combining a photo, 3s of sizzle audio, a 2s hot dog stand clip, and the word "lunch?" yielded "hot dog, 96.5%"
- Example request totaled 801 input tokens for $0.00012
- Clef also lowered Clef-flash pricing and boosted Clef inference speeds by up to 2.0x