Pruna AI launches P-Video-2: 20s 1080p/48fps video from text, image or audio, from $0.015/s
Scobleizer · x · 2026-09-11
Pruna AI announced P-Video-2, pitched as the fastest text/image/audio-to-video generation mode:
- Inputs: text, image, or audio-conditioned generation; model can also pick clip length from the prompt
- Output: up to 20s, up to 1080p at 24 or 48 fps
- Speed: draft from 0.41 s/s, standard from 0.91 s/s; draft mode offers faster previews at half price
- Pricing: draft from $0.015/s, standard from $0.025/s
- Controls: optional last-frame image, seed, aspect ratio, prompt upsampling
- Available via inference partners including Replicate, Cloudflare, RunPod, Segmind, eachlabs; benchmarks coming soon. A Pruna × Ideogram text-to-image playground also offers five thinking tiers trading speed/price/quality, 1K-2K output from $0.003/run, and structured JSON prompts for layout control
Related event: Pruna AI launches P-Video-2 for fast 20-second 1080p video generation(2 posts)→
More from Multimodal
- Voice-Cloned AI Singing Still Out of Reach: Testing LTX, MiniMax, AceStep — why_not_zoidberg_82 · 2026-09-11
- UMG and ElevenLabs sign multi-year licensing deal for licensed AI music platform — lukeharries · 2026-09-11
- Claude Opus 4.6 writes a full concept album: AI band 'The Assistant' lands on Spotify — ctjlewis · 2026-09-11
- MiniMax H3 video gen runs on an RTX 5050 in 5 minutes, may replace LoRAs — Creative_aidumpster · 2026-09-11
- Minimax H3 Is Fun — Full Video Workflow Shared as a Gist — Tryingmybestmama · 2026-09-11
- Minimax nails a foldable doodle in one shot, adds its own livestream background — Tokyo_Jab · 2026-09-11