the multi-silicon inference platform for video generation
London & SF
- Tweets 298
- Following 73
- Followers 2,680
- Likes 272
Pinned Tweet
In case you missed it, we teamed up with @tenstorrent and delivered realtime video. Video generated faster than it takes to play it!
If you want to test it, sign up at the link the comments.
7
14
57
343,471
Prodia retweeted
we just launched two new versions of minimax h3 that are dramatically cheaper than fal h3 max turbo at comparative quality 😎
4
3
34
1,515
Prodia retweeted
How to use this:
1. Look at your use case specific rankings
2. Pick a price point and pick 3 models to test in the spread below that price point
3. Crosscheck against the common pressure tests AA also now has rankings for and check those rankings for problems that are typical in your use case
Example:
use case = nail polish brand wants to make videos for 36 different skin tones
👇
retail & ecommerce use case —> stay below price point —> cross-ref vs the human anatomy rankings —> test, test, test
We are launching AA-Video-T2V v2.0, our new benchmark for evaluating text to video models, alongside AA-Video-T2V-Silent v2.0 for video generation without audio. Built on a new methodology, it judges every model at 1080p on a regularly refreshed prompt set, and ranks them across 10 use cases, 10 capabilities and a wide range of styles.
Video models are being adopted across more industries and workflows, from film studios to advertising agencies. Our new benchmark not only ranks models overall, but also shows which model is best for specific use case and video generation capability. Use cases are grounded in how consumers and enterprises use video generation. Capabilities draw on lab and academic research, and on how creators and businesses push video models today. We tag every prompt by use case (such as Live-Action Film and Marketing & Advertising) and by the capability it tests (such as Text Rendering and Audio Synchronization), and the overall benchmark samples evenly across both. Because of this, the overall ranking reflects a model's versatility across use cases and well-roundedness across capabilities. We also tag each prompt by visual style, such as photorealistic, 3D render, cartoon and anime, and hand-drawn illustration.
AI video is also moving onto bigger screens and into production, from microdramas to movie theaters, while low barrier to generate is resulting in a proliferation of low quality AI video content. The quality bar keeps rising, so we now judge every clip at 1080p and high bitrate.
We are launching AA-Video-T2V v2.0 with more than 68,000 high quality human preference votes from private evaluators based in US/UK over 1,000 prompts, and AA-Video-T2V-Silent v2.0 with more than 47,000 votes over 500 prompts.
Initial insights from an in-depth analysis of the 10 highest ranking models on the Artificial Analysis AA-Video-T2V v2.0 Leaderboard:
➤ Wan 3.0 ranks #1 overall and leads 10 of the 20 category boards, including Cartoon and Anime style and Animation & Gaming use case, at $12 per minute of video.
➤ Dreamina Seedance 2.5 ranks #2 and is the human performance specialist, #1 on both Human Anatomy and Dialogue & Lip Sync. At $34.12/min it is the most expensive model in the top 10.
➤ MiniMax H3 (768p) ranks #3, statistically tied with Seedance 2.5 at $4.80/min, about 1/7 of the price. It is also #1 on Text Rendering.
➤ FLUX 3 ranks #4, with its strongest results on Text Rendering (#3) and Dialogue & Lip Sync (#2).
➤ Gemini Omni Flash 1.1 ranks #5 and is the graphic 2D and audio specialist, #1 on UI/UX & Motion Design use case, Flat Design style and Audio Synchronization capability.
See below for the use case, capability and style breakdowns 🧵
1
2
5
321
H3 from @MiniMax_AI is live on Prodia! 🎬
Independently ranked on Artificial Analysis: #1 for video editing with audio, #2 for text-to-video with audio, #2 for image-to-video.
What everyone's shipping with it:
▸ On-screen text that stays stable across every frame — kinetic type moves instead of warping
▸ Audio + picture generated in one pass, in stereo. Lip sync holds across a full sequence, not just one beat
▸ Native audio in 11 languages, voiceover and on-screen text in the same generation
▸ Higher share of usable takes per generation → fewer rerolls, lower spend
2
1
2
523
I need a break from AI...
switching off from tech... by painting generated art (with Prodia) to paint by tokens? what?
London AI teaming up with @prodialabs & @UseCorgi
to have a chill Sunday afternoon painting in Corgi Cafe London first collab!
Whos coming?
3
3
22
3,166
We're super excited to offer P-Image-Ideogram on Prodia! Link in the comments.
Speeds as low as 0.4s & pricing as low as $0.003. Try it out now!
P-Image-Ideogram dominate the speed-quality and price-quality Pareto frontiers for image generation. It is the result of a unique collaboration with @ideogram_ai.
- Four modes (Very low, low, medium, high) for 1K-2K image generation.
- Optimal quality-efficiency with 0.4s-7.5s latency, and $0.003-$0.03 price.
- Structured JSON control & exact color control.
Available via our inference partners @Replicate @inference_sh @scenario_gg @wavespeed_ai @wiroai @magnific @prodialabs @togethercompute @runware @lovart_ai @ComfyUI @LeonardoAi @kittldesign @GammaApp @Picsart @TellersAI @Cloudflare Validated by our benchmark partners @DesignArena, @datapointai, @RapidataAI
👉 Try it on the playground for free: buff.ly/QmnA5P5
🧩 Sign in on the API: buff.ly/0iaZy8g
📚 Model page: buff.ly/XRaKSb0
1
2
7
904
Same 4-model video grid, different prompt. Wan 2.2, P-Video, Seedance Lite, Seedance Pro Turbo.
Variance between models on identical prompts is why multi-model access matters in production.
1
374
FLUX.2 [dev] vs. FLUX.2 [klein] 9B
4.5s at $12/1K vs 0.9s at $6/1K. 5x faster, half the price.
For product catalogs and e-commerce thumbnails, that math compounds fast.
2
223
FLUX.2 [klein] 4B. 100ms. Near-instant.
4B parameters doing work that used to take much larger models seconds. Reflections, neon glow, atmospheric depth. Fraction of the compute.
1
1
287
Same prompt, 4 models. $3 to $80 per 1K gens.
At 100K gens/month: klein 4B costs $300. Nano Banana 2 costs $8,000. Same prompt, massive cost gap.
2
2
270
FLUX.2 [flex] vs Nano Banana 2
12.7s at $60/1K vs 24.5s at $80/1K. Open source running 2x faster, 25% cheaper.
The quality speaks for itself.
2
266
Product shot: Recraft V4 vs Seedream 5 Lite.
Material rendering, lighting, composition. Both handle it, but reflections and textures differ noticeably.
At e-commerce scale, these nuances show up in conversions.
2
238
4 video models, 1 prompt.
Wan 2.2, P-Video, Seedance Lite, Seedance Pro Turbo. All different results.
Motion, consistency, speed, cost. Each has a sweet spot. Run them side by side to find yours.
2
3
338
Wan 2.2 Lightning generating video.
Open source video gen caught up faster than expected. Solid motion quality at a fraction of closed-source pricing.
2
214
Recraft V4 on text rendering. ~12s, $40/1K.
If your product needs readable text in generated images (menus, signs, packaging), this is where Recraft shines.
2
213
FLUX.2 [flex] vs Nano Banana 2, final round.
Across all five prompts: flex stays ahead on speed and cost. Optimized open source inference is pulling ahead.
1
4
189
Recraft V4 vs Seedream 5 Lite on a portrait.
~12s/$40 per 1K vs ~19s/$35 per 1K. Different approaches, different results. For user-facing portraits, these differences matter.
3
139
Prodia retweeted
@prodialabs - the fastest image and video generation API, one call to the best models
1
2
5
152