Pretty wild what 900M params can do – PaddleOCR VL > SOTA on OmniDocBench v1.0 & v1.5 (text, tables, formulas, charts, reading order)
> Multilingual – 109 languages (Latin, Arabic, Cyrillic, Devanagari, Thai, etc)
> Handles handwriting, historical docs, noisy scans
> Supports
TOOLS
-
PaddleOCR VL 900M Params Achieves SOTA on OmniDocBench
By
–
-
Photo AI Achieves Best Skin Texture With Clarity Upscaler
By
–
Photo AI now has the best skin texture of any image generator I think
— @levelsio (@levelsio) 17 octobre 2025
All thanks to @philz1337x's Clarity Upscaler! https://t.co/MDZ0mmmSGw pic.twitter.com/dbgAgNKwDIPhoto AI now has the best skin texture of any image generator I think All thanks to @philz1337x
's Clarity Upscaler! -
Poe App Creator Enhanced with Claude Code Integration
By
–
We made a major upgrade to Poe App Creator, and it is now built on top of Claude Code. This update will generally produce significantly better results, especially when building and iterating on complex applications inside of Poe. (1/2)
-
Poe App Creator Expands Access to 200+ AI Models
By
–
As usual, Poe apps you create with App Creator can build on top of any combination of the 200+ models on Poe. You can try the new experience at https://
poe.com/App-Creator, and across the Poe iOS, Android, Mac, and Windows apps. (2/2) -
Create seamless before-to-after transitions with first and last frame input
By
–
Use first/last frame input to create seamless before-to-after transitions, like for designing a room: pic.twitter.com/z1DR2arB4W
— Replicate (@replicate) 17 octobre 2025Use first/last frame input to create seamless before-to-after transitions, like for designing a room:
-
Create UGC-Style Videos with Reference Image Input
By
–
Create UGC-style videos with reference image input: pic.twitter.com/4pzi5ZP2xi
— Replicate (@replicate) 17 octobre 2025Create UGC-style videos with reference image input:
-
PhotoAI Face Transformation: Asian E-girl Style Generation
By
–
Pls ask PhotoAI to actually make an Asian e girl version of your real face I wanna see that so bad
-
Sora 2 Now Available on ChatLLM with Sound Features
By
–
Sora 2 is incredibly cool; it’s affordable and even features sound. It’s currently available on ChatLLM (as well as other video-generators, including the newly released Veo 3.1!)
-
AR and Generative AI Enable Expert-Level Work Without Training
By
–
La réalité augmentée + l'IA générative permettent à n'importe qui de réaliser un travail d'expert sans formation préalable.
— VISION IA (@vision_ia) 17 octobre 2025
C'est déjà une réalité. 😱 pic.twitter.com/RG9QXfRPj4Augmented reality + generative AI allow anyone to perform expert-level work without prior training. It's already a reality.
-
Adapting web app to mobile shopping with ChatGPT
By
–
Instead of rebuilding from scratch,
— God of Prompt (@godofprompt) 17 octobre 2025
I asked ChatGPT to take my Lovable web app prompt and Supabase schema
and adapt them for mobile shopping.
The spec came back with:
– Clean card layouts
– Bottom tab navigation (Home, Categories, Cart, Orders, Account)
– Carousels, loaders,… pic.twitter.com/eGgwUk0DDvInstead of rebuilding from scratch, I asked ChatGPT to take my Lovable web app prompt and Supabase schema and adapt them for mobile shopping. The spec came back with: – Clean card layouts
– Bottom tab navigation (Home, Categories, Cart, Orders, Account)
– Carousels, loaders,