Founder, President & Chief Scientist @PrunaAI Prev. @Twitter research, Ph.D. in ML @TU_Muenchen
Joined January 2019
- Tweets687
- Following187
- Followers755
- Likes1.8K
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
Below 4 bits, the โimplementation detailsโ become the method. Representation, calibration data and recovery data decide whether a 2-bit model works at all.
1๏ธโฃ Representation dominates: At ~2 bits, when scalar PTQ falls to chance, vector PTQ can maintain accuracy.
2๏ธโฃ Calibration and recovery data must be chosen jointly: Using recovery data different from calibration performs better.
3๏ธโฃ Recovery is cheap but model-dependent: while vector quantization recovery works well on some models with 1.4โ30x less compute than any QAT, no recovery worked best on some models.
๐ Paper: โBeyond 4 bits: Data-Centric Post-training Quantization and Recoveryโ
๐ฉโ๐ฌ Authors: @RheaSukthanker, @riccardograzzi, @jameshensman, @CameronPashmina
๐ Link: arxiv.org/abs/2610.28537
Bertrand Charpentier retweeted
This week we're reviewing your ๐ฃ๐ฟ๐๐ป๐ฎ ๐๐ฟ๐ฒ๐ฎ๐๐ถ๐๐ฒ ๐ฃ๐ฟ๐ผ๐ด๐ฟ๐ฎ๐บ โจ applications! Hurry up and apply now to be considered!
- ๐ช๐ต๐ฎ๐โ๐ ๐ถ๐ ๐ฎ๐ฏ๐ผ๐๐? A space for AI artists and creators to experiment with faster, more affordable, and controllable models.
- ๐ช๐ต๐ผโ๐ ๐ถ๐ ๐ณ๐ผ๐ฟ? The program is open to artists, filmmakers, animators, designers, developers, creative technologists, and AI-native creators.
- ๐ช๐ต๐ฎ๐ ๐ฑ๐ผ ๐๐ผ๐ ๐ด๐ฒ๐? Become an ๐๐
๐ฝ๐น๐ผ๐ฟ๐ฒ๐ฟ or a ๐๐ฒ๐น๐น๐ผ๐ and get supported with monthly credits, early model access, monthly paid campaigns, and opportunities to be featured.
Just original concepts. Share your style with creative freedom and hundreds of possibilities.
Read more information and apply here: pruna.ai/blog/pruna-creativeโฆ
Bertrand Charpentier retweeted
First test of P-Video-2-Pro-Reference, using my character sheet as a visual reference for a 15-second, 16:9 fashion clip.
Fun to see a character design move from a static reference into video. Looking forward to trying more reference combinations. @PrunaAI
Introducing P-Video-2-Pro-Reference based on @MiniMax_AI H3! ๐ฌ
Combine images, videos, and audio references to guide how your video looks, moves, and sounds. Keep characters consistent, transfer motion, and bring different visual elements together in a single generation.
โข ๐๐ป๐ฝ๐๐: Text prompt with image and/or video references, plus optional audio
โข ๐ฅ๐ฒ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ๐: Up to 9 images, 3 videos, and 3 audio files
โข ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป: 5โ15s at 480p or 768p
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Output starting at $0.015/s in Cost mode, with reference inputs billed separately
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: First and last frames, seed, aspect ratio, and prompt upsampling
Two modes: Speed and Cost. Choose the trade-off that fits your workflow.
Available via our inference partners @Cloudflare, @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, @inference_sh, @_segmind, @togethercompute, @EdenAIco, and @runware
Validated by our benchmark partners @datapointai, @RapidataAI, and @DesignArena
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ API: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
Munich, join us for Hacktoberfest with @MLHacks !
We're hosting a casual meetup at the Pruna office on October 29!
Come grab a pizza, have a drink, meet other developers, and chat about open source, side projects, or whatever you're building.
No talks or pitches, just a relaxed evening with the local community.
๐ Pruna office, Munich
๐
October 29, 6โ9 PM
Hope to see you there!
๐ Sign up: events.mlh.com/events/15374-โฆ
Bertrand Charpentier retweeted
AI VIDEO STILL LOSES YOUR CHARACTER BETWEEN SHOTS.
@PrunaAI released P-Video-2-Pro Reference for that. you give it up to 9 images, 3 videos and 3 audio files as references, and it keeps the subject consistent through the clip. built on MiniMax H3.
it can also copy the motion from a reference video, or mix elements from different references in one generation.
output starts at $0.015 per second in cost mode, with references billed on top. the API is live now.
๐ค Paid partnership
Introducing P-Video-2-Pro-Reference based on @MiniMax_AI H3! ๐ฌ
Combine images, videos, and audio references to guide how your video looks, moves, and sounds. Keep characters consistent, transfer motion, and bring different visual elements together in a single generation.
โข ๐๐ป๐ฝ๐๐: Text prompt with image and/or video references, plus optional audio
โข ๐ฅ๐ฒ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ๐: Up to 9 images, 3 videos, and 3 audio files
โข ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป: 5โ15s at 480p or 768p
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Output starting at $0.015/s in Cost mode, with reference inputs billed separately
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: First and last frames, seed, aspect ratio, and prompt upsampling
Two modes: Speed and Cost. Choose the trade-off that fits your workflow.
Available via our inference partners @Cloudflare, @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, @inference_sh, @_segmind, @togethercompute, @EdenAIco, and @runware
Validated by our benchmark partners @datapointai, @RapidataAI, and @DesignArena
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ API: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
p-video-2-pro-reference by @PrunaAI is now available on Design Arena!
Built on @MiniMax_AI H3, p-video-2-pro-reference combines image, video, and audio references to give creators greater control over video generation. It supports consistent characters, motion transfer, and multimodal inputs, with adjustable duration, aspect ratio, and first and last frames.
Congrats to the @PrunaAI team on the launch!
Introducing P-Video-2-Pro-Reference based on @MiniMax_AI H3! ๐ฌ
Combine images, videos, and audio references to guide how your video looks, moves, and sounds. Keep characters consistent, transfer motion, and bring different visual elements together in a single generation.
โข ๐๐ป๐ฝ๐๐: Text prompt with image and/or video references, plus optional audio
โข ๐ฅ๐ฒ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ๐: Up to 9 images, 3 videos, and 3 audio files
โข ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป: 5โ15s at 480p or 768p
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Output starting at $0.015/s in Cost mode, with reference inputs billed separately
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: First and last frames, seed, aspect ratio, and prompt upsampling
Two modes: Speed and Cost. Choose the trade-off that fits your workflow.
Available via our inference partners @Cloudflare, @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, @inference_sh, @_segmind, @togethercompute, @EdenAIco, and @runware
Validated by our benchmark partners @datapointai, @RapidataAI, and @DesignArena
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ API: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
ai video was never the hard part
keeping your product looking the same in every shot was
@prunaai just dropped p-video-2-pro reference and it locks your product, character or voice straight from your refs
up to 9 images, 3 videos and 3 audio clips, sound included
from 2 cents a second
consistent ai video just got cheap
API๐dashboard.pruna.ai/login
๐ค Paid partnership
Introducing P-Video-2-Pro-Reference based on @MiniMax_AI H3! ๐ฌ
Combine images, videos, and audio references to guide how your video looks, moves, and sounds. Keep characters consistent, transfer motion, and bring different visual elements together in a single generation.
โข ๐๐ป๐ฝ๐๐: Text prompt with image and/or video references, plus optional audio
โข ๐ฅ๐ฒ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ๐: Up to 9 images, 3 videos, and 3 audio files
โข ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป: 5โ15s at 480p or 768p
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Output starting at $0.015/s in Cost mode, with reference inputs billed separately
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: First and last frames, seed, aspect ratio, and prompt upsampling
Two modes: Speed and Cost. Choose the trade-off that fits your workflow.
Available via our inference partners @Cloudflare, @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, @inference_sh, @_segmind, @togethercompute, @EdenAIco, and @runware
Validated by our benchmark partners @datapointai, @RapidataAI, and @DesignArena
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ API: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
PRUNA JUST DROPPED THE MOST CONSISTENT AI VIDEO MODEL I'VE TESTED AND I'M GOING TO EXPLAIN WHY.
Same surfer. Same board. Every single angle.
This is P-Video-2-Pro Reference by @PrunaAI.
References โ Prompt โ Locked Subject โ Video With Sound.
โ You send up to 9 images, 3 videos and 3 audio clips.
โ It locks your subject: a character, a product, a clip or a voice.
โ It returns an MP4 at 24 fps with generated sound.
โ Clips go from 5 to 15 seconds at 480p or 768p.
โ Built on MiniMax H3 Ref2V.
Consistency was always the weak spot of AI video. Not anymore.
Introducing P-Video-2-Pro-Reference based on @MiniMax_AI H3! ๐ฌ
Combine images, videos, and audio references to guide how your video looks, moves, and sounds. Keep characters consistent, transfer motion, and bring different visual elements together in a single generation.
โข ๐๐ป๐ฝ๐๐: Text prompt with image and/or video references, plus optional audio
โข ๐ฅ๐ฒ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ๐: Up to 9 images, 3 videos, and 3 audio files
โข ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป: 5โ15s at 480p or 768p
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Output starting at $0.015/s in Cost mode, with reference inputs billed separately
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: First and last frames, seed, aspect ratio, and prompt upsampling
Two modes: Speed and Cost. Choose the trade-off that fits your workflow.
Available via our inference partners @Cloudflare, @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, @inference_sh, @_segmind, @togethercompute, @EdenAIco, and @runware
Validated by our benchmark partners @datapointai, @RapidataAI, and @DesignArena
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ API: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
AI VIDEO IS MOVING BEYOND TEXT-TO-VIDEO
Generating a clip is one thing. Getting the exact character, product, motion and sound you want is another.
P-Video-2-Pro Reference by @PrunaAI is built for that:
โ Combine a prompt with up to 9 images, 3 videos and 3 audio clips
โ Keep subjects consistent using visual and audio references
โ Generate 5โ15 second clips with sound
โ Choose between speed and cost modes, starting at $0.02/s
Instead of describing every detail from scratch, you can give the model references for what you actually want it to create.
More control over the output, less guesswork.
Introducing P-Video-2-Pro-Reference based on @MiniMax_AI H3! ๐ฌ
Combine images, videos, and audio references to guide how your video looks, moves, and sounds. Keep characters consistent, transfer motion, and bring different visual elements together in a single generation.
โข ๐๐ป๐ฝ๐๐: Text prompt with image and/or video references, plus optional audio
โข ๐ฅ๐ฒ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ๐: Up to 9 images, 3 videos, and 3 audio files
โข ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป: 5โ15s at 480p or 768p
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Output starting at $0.015/s in Cost mode, with reference inputs billed separately
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: First and last frames, seed, aspect ratio, and prompt upsampling
Two modes: Speed and Cost. Choose the trade-off that fits your workflow.
Available via our inference partners @Cloudflare, @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, @inference_sh, @_segmind, @togethercompute, @EdenAIco, and @runware
Validated by our benchmark partners @datapointai, @RapidataAI, and @DesignArena
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ API: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
Video models give you creative freedom, but you lose control of the subject.
P-Video-2-Pro Reference does the opposite:
โ Lock in your subject with up to 9 reference images, 3 videos, and 3 audio clips
โ Generates 5โ15s clips at 480p or 768p, 24fps, with sound included
โ Built on MiniMax H3 Ref2V
โ From $0.02/s in cost mode, from 1.9s generation time in speed mode
โ Available via API right now
The same character, product, or voice. In every clip.
@PrunaAI
๐ค Paid partnership
Introducing P-Video-2-Pro-Reference based on @MiniMax_AI H3! ๐ฌ
Combine images, videos, and audio references to guide how your video looks, moves, and sounds. Keep characters consistent, transfer motion, and bring different visual elements together in a single generation.
โข ๐๐ป๐ฝ๐๐: Text prompt with image and/or video references, plus optional audio
โข ๐ฅ๐ฒ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ๐: Up to 9 images, 3 videos, and 3 audio files
โข ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป: 5โ15s at 480p or 768p
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Output starting at $0.015/s in Cost mode, with reference inputs billed separately
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: First and last frames, seed, aspect ratio, and prompt upsampling
Two modes: Speed and Cost. Choose the trade-off that fits your workflow.
Available via our inference partners @Cloudflare, @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, @inference_sh, @_segmind, @togethercompute, @EdenAIco, and @runware
Validated by our benchmark partners @datapointai, @RapidataAI, and @DesignArena
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ API: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
THE GUY ISNโT REAL. THE SKY ISNโT REAL. THE SOUND ISNโT REAL. AND GENERATING IT COST ME LESS THAN A COFFEE.
one API call. thatโs it.
this is @PrunaAI a new reference-to-video model built on MiniMax H3.
itโs not regular text-to-video. you donโt describe something and hope for the best.
โ you feed it photos or clips of your subject
โ give it a prompt
โ and it keeps that exact subject consistent throughout the whole video
it accepts up to 9 images, 3 videos and 3 audio clips as reference. returns an mp4 at 24fps with generated sound.
โ 5-15 seconds
โ 480p or 768p
โ from $0.02/s
subject consistency is the part most video models still get wrong. you feed it a face and 3 seconds later itโs a different person.
here the subject stays locked. frame by frame.
the API is live.
๐ค Paid partnership
Introducing P-Video-2-Pro-Reference based on @MiniMax_AI H3! ๐ฌ
Combine images, videos, and audio references to guide how your video looks, moves, and sounds. Keep characters consistent, transfer motion, and bring different visual elements together in a single generation.
โข ๐๐ป๐ฝ๐๐: Text prompt with image and/or video references, plus optional audio
โข ๐ฅ๐ฒ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ๐: Up to 9 images, 3 videos, and 3 audio files
โข ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป: 5โ15s at 480p or 768p
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Output starting at $0.015/s in Cost mode, with reference inputs billed separately
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: First and last frames, seed, aspect ratio, and prompt upsampling
Two modes: Speed and Cost. Choose the trade-off that fits your workflow.
Available via our inference partners @Cloudflare, @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, @inference_sh, @_segmind, @togethercompute, @EdenAIco, and @runware
Validated by our benchmark partners @datapointai, @RapidataAI, and @DesignArena
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ API: dashboard.pruna.ai/login
Bertrand Charpentier retweeted
@PrunaAI's P-Video-2 Pro now takes image, video, and audio references ๐ทโ
The same model, integration, now accepting references alongside your prompt to condition the generation.
- referenceImages, referenceVideos, referenceAudios
- Works alongside first and last frame control
- First audio input on P-Video-2 Pro
- 5-15s clips, up to 768p, with native audio
From $0.01/s with a 50% launch discount through October 29, on Runware.
Introducing P-Video-2-Pro-Reference based on @MiniMax_AI H3! ๐ฌ
Combine images, videos, and audio references to guide how your video looks, moves, and sounds. Keep characters consistent, transfer motion, and bring different visual elements together in a single generation.
โข ๐๐ป๐ฝ๐๐: Text prompt with image and/or video references, plus optional audio
โข ๐ฅ๐ฒ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ๐: Up to 9 images, 3 videos, and 3 audio files
โข ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป: 5โ15s at 480p or 768p
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Output starting at $0.015/s in Cost mode, with reference inputs billed separately
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: First and last frames, seed, aspect ratio, and prompt upsampling
Two modes: Speed and Cost. Choose the trade-off that fits your workflow.
Available via our inference partners @Cloudflare, @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, @inference_sh, @_segmind, @togethercompute, @EdenAIco, and @runware
Validated by our benchmark partners @datapointai, @RapidataAI, and @DesignArena
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ API: dashboard.pruna.ai/login
Welcome to P-Video-2-Pro-Reference, for full control over video generation at maximum speed and the lowest price.
- Full control: 9 reference images, 3 reference videos, 3 reference audio clips.
- Best speed: starting at $0.015 per second of output video.
- Best price: a 5-second clip ready in 1.9 seconds.
Try it on dashboard.pruna.ai/login
What do you want next?๐
Introducing P-Video-2-Pro-Reference based on @MiniMax_AI H3! ๐ฌ
Combine images, videos, and audio references to guide how your video looks, moves, and sounds. Keep characters consistent, transfer motion, and bring different visual elements together in a single generation.
โข ๐๐ป๐ฝ๐๐: Text prompt with image and/or video references, plus optional audio
โข ๐ฅ๐ฒ๐ณ๐ฒ๐ฟ๐ฒ๐ป๐ฐ๐ฒ๐: Up to 9 images, 3 videos, and 3 audio files
โข ๐๐๐ฟ๐ฎ๐๐ถ๐ผ๐ป: 5โ15s at 480p or 768p
โข ๐ฃ๐ฟ๐ถ๐ฐ๐ถ๐ป๐ด: Output starting at $0.015/s in Cost mode, with reference inputs billed separately
โข ๐๐ผ๐ป๐๐ฟ๐ผ๐น: First and last frames, seed, aspect ratio, and prompt upsampling
Two modes: Speed and Cost. Choose the trade-off that fits your workflow.
Available via our inference partners @Cloudflare, @scenario_gg, @replicate, @WiroAI, @TellersAI, @eachlabs, @wavespeed_ai, @inference_sh, @_segmind, @togethercompute, @EdenAIco, and @runware
Validated by our benchmark partners @datapointai, @RapidataAI, and @DesignArena
๐ Model docs: docs.api.pruna.ai/guides/modโฆ
๐งฉ API: dashboard.pruna.ai/login
No vague posting: #1 in quality, speed, and price.
Image-to-Video: P-Video-2-Pro is the #1 ELO model on @DesignArena.
โข Speed: #1 ยท 1328 Elo
โข Quality: #2 ยท 1326 Elo
โข Cost: #3 ยท 1324 Elo
One model, three modes. Depending on whether you want to optimize for cost, speed, or quality.
Benchmark by @DesignArena.
๐ Documentation: buff.ly/bHQHKV0
๐งฉ API: dashboard.pruna.ai
๐ฅ P-Bench: buff.ly/81e3AuS
Bertrand Charpentier retweeted
The Pruna community used P-Video-Edit to change wardrobes, products, seasons, locations, characters, and entire worlds while keeping the original motion.
Don't reshoot it, edit it.
Thank you all of you for making these creations with P-Video-Edit!
๐น Watch the videos here: youtube.com/@PrunaAI/playlisโฆ
๐ Documentation: buff.ly/0P6Derc
๐งฉ API: buff.ly/1l6X0XB
Bertrand Charpentier retweeted
Text-to-Video: P-Video-2-Pro is now one of the strongest model families on the @DesignArena leaderboard.
Three P-Video-2-Pro modes rank near the top:
โข Cost: #4 ยท 1321 Elo
โข Quality: #5 ยท 1304 Elo
โข Speed: #7 ยท 1294 Elo
And on the Preference ร Speed benchmark, Pruna makes up 4 of the 5 models on the high-preference section of the Pareto frontier.
All three P-Video-2-Pro modes sit on the frontier, offering different trade-offs between preference and generation time.
Cost also lands on the Preference ร Price Pareto frontier, starting at just $0.01/s.
One model, three modes. Depending on whether you want to optimise for cost, speed, or quality.
Benchmark by @DesignArena
๐ Documentation: docs.api.pruna.ai/guides/modโฆ
๐งฉ API: dashboard.pruna.ai/login
๐ฅ P-Bench: huggingface.co/spaces/PrunaAโฆ
Bertrand Charpentier retweeted
Want Pruna news and product updates in your inbox?
Join our update list to follow what we're building at Pruna.
Open the contact page, add your details, tick "Keep me updated on Pruna news and product updates", and hit Submit.
๐ pruna.ai/contact
Bertrand Charpentier retweeted
There are trips that feel like we've all dreamed of, even though we've never been. Fields of flowers, silent mountains, and a figure that keeps walkingโI want this world to feel warm, calm, and a little unreal.
Made with P-Video-2-Pro from @PrunaAI
prompt ๐