Founder, President & Chief Scientist @PrunaAI Prev. @Twitter research, Ph.D. in ML @TU_Muenchen
Joined January 2019
- Tweets722
- Following189
- Followers768
- Likes1.9K
Pinned Tweet
Today, we are happy to release P Video 2 Pro, Based on H3 from @MiniMax_AI .
Highest Quality, Highest Speed.
More benchmarks (and more models?!) coming soon!
Today, we launch P-Video-2-Pro, based on MiniMax H3: The best quality. The best speed.
Text or image in. Video and audio out.
โข ๐๐ป๐ฝ๐๐: Text or first-frame image, with optional last-frame conditioning
โข ๐ฆ๐ฝ๐ฒ๐ฒ๐ฑ: ~2.0s for a 5s clip at 480p ยท ~4.3s at 768p in Speed mode
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: From $0.02/s Speed ยท $0.04/s Quality
โข ๐ ๐ผ๐ฑ๐ฒ๐: Speed or Quality
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: First and last frame, seed, aspect ratio, and prompt upsampling with Off, Turbo, or Max
Available via our inference partners @replicate, @Cloudflare, @eachlabs, @scenario_gg, @TellersAI, @wavespeed_ai, @wiroai, @runware, and @togethercompute
Validated by our benchmark partners @datapointai, @RapidataAI, and @DesignArena. Benchmarks to be released soon ๐
๐ Free Playground: playground.pruna.ai/
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ API: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
We put P-Video-Upscale to the test!
In our own benchmarks against Topaz Starlight, Topaz Proteus, and FLUX.3 Upscaler, P-Video-Upscale is up to 11ร faster and 5ร cheaper at 1080p.
Here's how they compare๐
Introducing P-Video-Upscale! ๐ฌ
Fast video upscaling to 720p, 1080p, 2K, or 4K, while preserving the original aspect ratio and audio. Need smoother motion? Increase the frame rate to up to 60 fps.
And for launch, it's 50% off for the first two weeks, until November 12 at 18:00 CET!
Here's what you need to know:
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Starting from $0.01/s at 720p
โข ๐ฆ๐ฝ๐ฒ๐ฒ๐ฑ: ~1.7s per second of output video, depending on resolution
โข ๐๐ฟ๐ฎ๐บ๐ฒ ๐ฟ๐ฎ๐๐ฒ: Keep the original or interpolate up to 60 fps
โข ๐ง๐๐ฟ๐ฏ๐ผ ๐บ๐ผ๐ฑ๐ฒ: Faster processing with a slight quality trade-off
No need to regenerate a video just to get it in higher resolution.
Available via our inference partners @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, Sociaro, and @runware.
Benchmarks by our benchmark partners @datapointai and @RapidataAI are coming soon.
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ Try the model through the API or Playground: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
No need to regenerate a video just to increase its resolution.
P-Video-Upscale by @PrunaAI can upscale to 4K while preserving aspect ratio and audio, with frame interpolation up to 60 fps.
Starts at $0.01/s for 720p; get 50% off until Nov 12, 18:00 CET.
Introducing P-Video-Upscale! ๐ฌ
Fast video upscaling to 720p, 1080p, 2K, or 4K, while preserving the original aspect ratio and audio. Need smoother motion? Increase the frame rate to up to 60 fps.
And for launch, it's 50% off for the first two weeks, until November 12 at 18:00 CET!
Here's what you need to know:
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Starting from $0.01/s at 720p
โข ๐ฆ๐ฝ๐ฒ๐ฒ๐ฑ: ~1.7s per second of output video, depending on resolution
โข ๐๐ฟ๐ฎ๐บ๐ฒ ๐ฟ๐ฎ๐๐ฒ: Keep the original or interpolate up to 60 fps
โข ๐ง๐๐ฟ๐ฏ๐ผ ๐บ๐ผ๐ฑ๐ฒ: Faster processing with a slight quality trade-off
No need to regenerate a video just to get it in higher resolution.
Available via our inference partners @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, Sociaro, and @runware.
Benchmarks by our benchmark partners @datapointai and @RapidataAI are coming soon.
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ Try the model through the API or Playground: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
Stop regenerating video just to get a higher resolution.
Upgrade to 720p, 1080p, 2K, or 4K, while maintaining audio and aspect ratio. Up to 60 fps for smoother motion.
Starting at $0.01/second. 50% off until November 12th, 6:00 PM CET.
@PrunaAI
Introducing P-Video-Upscale! ๐ฌ
Fast video upscaling to 720p, 1080p, 2K, or 4K, while preserving the original aspect ratio and audio. Need smoother motion? Increase the frame rate to up to 60 fps.
And for launch, it's 50% off for the first two weeks, until November 12 at 18:00 CET!
Here's what you need to know:
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Starting from $0.01/s at 720p
โข ๐ฆ๐ฝ๐ฒ๐ฒ๐ฑ: ~1.7s per second of output video, depending on resolution
โข ๐๐ฟ๐ฎ๐บ๐ฒ ๐ฟ๐ฎ๐๐ฒ: Keep the original or interpolate up to 60 fps
โข ๐ง๐๐ฟ๐ฏ๐ผ ๐บ๐ผ๐ฑ๐ฒ: Faster processing with a slight quality trade-off
No need to regenerate a video just to get it in higher resolution.
Available via our inference partners @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, Sociaro, and @runware.
Benchmarks by our benchmark partners @datapointai and @RapidataAI are coming soon.
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ Try the model through the API or Playground: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
No need to regenerate. Just upscale!
P-Video-Upscale takes your videos up to 4K, preserves the original aspect ratio and audio, and boosts frame rates up to 60 FPS. All at impressive speed and launch pricing starting at just $0.01/s.
Big upgrade for AI video creators!
@PrunaAI
๐ค Made with AI
Introducing P-Video-Upscale! ๐ฌ
Fast video upscaling to 720p, 1080p, 2K, or 4K, while preserving the original aspect ratio and audio. Need smoother motion? Increase the frame rate to up to 60 fps.
And for launch, it's 50% off for the first two weeks, until November 12 at 18:00 CET!
Here's what you need to know:
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Starting from $0.01/s at 720p
โข ๐ฆ๐ฝ๐ฒ๐ฒ๐ฑ: ~1.7s per second of output video, depending on resolution
โข ๐๐ฟ๐ฎ๐บ๐ฒ ๐ฟ๐ฎ๐๐ฒ: Keep the original or interpolate up to 60 fps
โข ๐ง๐๐ฟ๐ฏ๐ผ ๐บ๐ผ๐ฑ๐ฒ: Faster processing with a slight quality trade-off
No need to regenerate a video just to get it in higher resolution.
Available via our inference partners @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, Sociaro, and @runware.
Benchmarks by our benchmark partners @datapointai and @RapidataAI are coming soon.
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ Try the model through the API or Playground: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
Upscaling videos is finally cheap. And of course, itโs already on @TellersAI .
Congrats @PrunaAI! ๐
This entire video was made with one prompt in Tellers: generation, titles, upscaling, masking, compositing, VFX, and voice.
Make people happy: upscale your videos to 4K in seconds. โจ
Introducing P-Video-Upscale! ๐ฌ
Fast video upscaling to 720p, 1080p, 2K, or 4K, while preserving the original aspect ratio and audio. Need smoother motion? Increase the frame rate to up to 60 fps.
And for launch, it's 50% off for the first two weeks, until November 12 at 18:00 CET!
Here's what you need to know:
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Starting from $0.01/s at 720p
โข ๐ฆ๐ฝ๐ฒ๐ฒ๐ฑ: ~1.7s per second of output video, depending on resolution
โข ๐๐ฟ๐ฎ๐บ๐ฒ ๐ฟ๐ฎ๐๐ฒ: Keep the original or interpolate up to 60 fps
โข ๐ง๐๐ฟ๐ฏ๐ผ ๐บ๐ผ๐ฑ๐ฒ: Faster processing with a slight quality trade-off
No need to regenerate a video just to get it in higher resolution.
Available via our inference partners @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, Sociaro, and @runware.
Benchmarks by our benchmark partners @datapointai and @RapidataAI are coming soon.
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ Try the model through the API or Playground: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
P-Q-Image-2.1 brings Qwen-Image-2.1 generation and editing down to six steps, with no CFG and up to 6.3ร faster generation.
๐ค modelscope.ai/models/PrunaAIโฆ
โก Six steps, same workflow: The v0.2 LoRA adapters run on top of Qwen-Image-2.1 while keeping its base pipeline, text encoder, and VAE unchanged.
๐ผ๏ธ Generate or edit: Use one adapter for text-to-image generation and single- or multi-image editing, with up to three reference images.
โจ Skip the separate prompt-upscaling pass: The 6-steps-upsampled variant incorporates prompt-upscaling benefits into the adapter, reducing end-to-end generation time from a short prompt.
๐๏ธ Choose the trade-off: Use 6-steps for behavior closer to the base model, or 6-steps-upsampled when prioritizing visual quality from concise prompts.
๐ ๏ธ Built for the Qwen stack: Start at 1024 ร 1024 and use the matching six-step schedule. Higher resolutions may work, but were outside the adapterโs training coverage.
๐ Qwen Research License Agreement. This is a LoRA adapter, not a standalone model; it requires the Qwen-Image-2.1 base weights.
Bertrand Charpentier retweeted
Talking-avatar videos are getting faster and more affordable. According to @PrunaAI benchmark, P-Video-2-Pro-Avatar ranked #1 in overall human evaluation at 768p, runs up to 3ร faster, and costs 7ร less at 480p.
Starting at $0.02 per second. Ideal for UGC, ads, and music videos.
P-Video-2-Pro-Avatar is #1 in overall human ratings, at a fraction of the cost and in seconds.
- Quality: #1 of 10 models at 768p
- Cost: 7ร cheaper, just $0.02/s vs $0.14/s at 480p
- Speed: 3ร faster, just ~9.7s for a 10s clip at 480p
- # 1 for music videos at 768p and #2 at 480p
Across both Quality ร Price and Quality ร Speed, P-Video-2-Pro-Avatar stands out in the high-quality, low-cost, fast-generation region.
From UGC ads to music videos, create high-quality talking avatars without compromising on speed or budget.
Benchmarked by @RapidataAI.
๐งฉ API & Playground: buff.ly/OHtx3eM
๐ Documentation: buff.ly/bHQHKV0
๐ P-Bench: buff.ly/Fk54Z2K
Weโre open-sourcing more distillation work, this time including prompt upsampling distillation. ๐
End to end, from the raw prompt to the final image: 51s โ 1.2s i.e. ~42ร faster
Prompt upsampling used to be the bottleneck. Not anymore.
Today, we open-source v0.2 of P-Q-Image-2.1, two new 6-step LoRA adapters for Qwen-Image-2.1 by @Alibaba_Qwen. The first distillation that removes the prompt-upsampling step!
- ๐๐ผ๐ ๐ฒ๐ณ๐ณ๐ถ๐ฐ๐ถ๐ฒ๐ป๐ ๐ถ๐ ๐ถ๐? Generate or edit images in just 6 steps instead of 40. Use 6-steps for preserving inputs when editing and 6-steps-upsampled for improved visual quality and shorter end-to-end generation.
- ๐๐ผ๐ ๐ด๐ผ๐ผ๐ฑ ๐ถ๐ ๐ถ๐? 6-step distillation provides improved quality with few generation steps. 6-steps-upsampled is the first distillation that removes the need for a separate prompt-upsampling step. This means 41 to 80 seconds saved, depending on your prompt style.
- ๐๐ผ๐ ๐ฑ๐ผ๐ฒ๐ ๐ถ๐ ๐๐ผ๐ฟ๐ธ? The LoRA adapters load directly on top of Qwen-Image-2.1, so the pipeline stays unchanged. Few-step distillation cuts the diffusion cost, and the upsampled adapter also removes the prompt-upsampling step before image generation.
๐ค Open Weights: huggingface.co/PrunaAI/P-Q-Iโฆ
๐ HF Space: huggingface.co/spaces/PrunaAโฆ
A single avatar model is #1 on quality, speed, and price: P-Video-2-Pro-Avatar.
P-Video-2-Pro-Avatar is #1 in overall human ratings, at a fraction of the cost and in seconds.
- Quality: #1 of 10 models at 768p
- Cost: 7ร cheaper, just $0.02/s vs $0.14/s at 480p
- Speed: 3ร faster, just ~9.7s for a 10s clip at 480p
- # 1 for music videos at 768p and #2 at 480p
Across both Quality ร Price and Quality ร Speed, P-Video-2-Pro-Avatar stands out in the high-quality, low-cost, fast-generation region.
From UGC ads to music videos, create high-quality talking avatars without compromising on speed or budget.
Benchmarked by @RapidataAI.
๐งฉ API & Playground: buff.ly/OHtx3eM
๐ Documentation: buff.ly/bHQHKV0
๐ P-Bench: buff.ly/Fk54Z2K
Bertrand Charpentier retweeted
One portrait + one script = a talking video with natural expressions and accurate lip sync. No camera, no studio.
@PrunaAI P-Video-2-Pro-Avatar: from $0.02/s, 50% off until Nov 5, 18:00 CET.
Full details ๐
Introducing P-Video-2-Pro-Avatar, based on @MiniMax_AI H3 !
Turn a single portrait into a realistic talking video with natural expressions and accurate lip sync.
And for launch, it's 50% off for the first two weeks, until November 5 at 18:00 CET!
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: $0.02/s at 480p ยท $0.04/s at 768p
โข ๐ฆ๐ฝ๐ฒ๐ฒ๐ฑ: ~0.97 s at 480p ยท ~2.2 s at 768p per second of output video
โข ๐๐ป๐ฝ๐๐: One portrait with a text script or uploaded audio
โข ๐ฉ๐ผ๐ถ๐ฐ๐ฒ๐: 30 voices across 10 languages
โข ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป: 1sโ3min at 24 fps, in 480p or 768p
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: Voice delivery, scene motion, seed, and prompt upsampling
One image, two ways to bring it to life. Use your own audio or generate speech directly from text.
Available via our inference partners @replicate, @eachlabs, @Cloudflare, @scenario_gg, @wavespeed_ai, @WiroAI, Sociaro, @TellersAI, and @togethercompute.
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ Try the model through the API or Playground: dashboard.pruna.ai/login
๐๏ธ Last week's efficiency papers (1) highlighted that the biggest wins come from skipping work the model doesn't need and (2) that a speedup only exists on a given machine.
Here is my selection of 5 ML efficiency papers ๐งต
4๏ธโฃ Layers and diffusion steps of a video model have different sensitivities for final output quality. Allocate different attention sparsity at every layer/step to accelerate generation without impacting quality.
arxiv.org/abs/2610.29441
โญ Pick of the week: Joules-to-GDP precisely measures economic value per joule on 220 real professional tasks. They cover:
- Net value = outputโs dollar value โ AI cost (only if a blinded judge says the AI matches the expert).
- Time saving = expert hours โ AI hours on those wins.
E.g. on that yardstick, Gemma 4 31B delivers ~2.5ร the net $/joule of GLM 5.2.
arxiv.org/abs/2610.30858
Bertrand Charpentier retweeted
Tried the new talking portrait model and ended up with something that felt more like a short memory.
Same person, same soft voice shown in different styles and ages.
The expressions and lip sync is ๐ฅ
Made with the new P-Video-2-Pro-Avatar on @PrunaAI.
You can now play with all the fastest models on one interactive playground!
๐๐ก๐ ๐๐ฅ๐๐ฒ๐ ๐ซ๐จ๐ฎ๐ง๐ ๐ข๐ฌ ๐๐๐๐ค ๐๐ง๐ ๐๐๐ญ๐ญ๐๐ซ ๐ญ๐ก๐๐ง ๐๐ฏ๐๐ซ!
If you havenโt noticed yet, the Playground is available again, now powered directly by the API. This means:
1) Forget about sharing it with others and being limited
2) No more 5-generation limit -> Use your credits to create as much as you want.
Sign up now in the Dashboard, add your payment information, and get $20 in free credits (+100 video generations), with no other commitment required!
Start creating in the Playground today: buff.ly/OHtx3eM
Bertrand Charpentier retweeted
Pruna P-Video-2-Pro Avatar will be available in ComfyUI!
Introducing P-Video-2-Pro-Avatar, based on @MiniMax_AI H3 !
Turn a single portrait into a realistic talking video with natural expressions and accurate lip sync.
And for launch, it's 50% off for the first two weeks, until November 5 at 18:00 CET!
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: $0.02/s at 480p ยท $0.04/s at 768p
โข ๐ฆ๐ฝ๐ฒ๐ฒ๐ฑ: ~0.97 s at 480p ยท ~2.2 s at 768p per second of output video
โข ๐๐ป๐ฝ๐๐: One portrait with a text script or uploaded audio
โข ๐ฉ๐ผ๐ถ๐ฐ๐ฒ๐: 30 voices across 10 languages
โข ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป: 1sโ3min at 24 fps, in 480p or 768p
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: Voice delivery, scene motion, seed, and prompt upsampling
One image, two ways to bring it to life. Use your own audio or generate speech directly from text.
Available via our inference partners @replicate, @eachlabs, @Cloudflare, @scenario_gg, @wavespeed_ai, @WiroAI, Sociaro, @TellersAI, and @togethercompute.
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ Try the model through the API or Playground: dashboard.pruna.ai/login