I’m in Asia right now, on the opposite schedule of almost everyone else I know using Claude Code and Codex. And I have to say, I’ve been getting way better results than when I’m in the States. Time-of-day-driven performance changes are real.
GENERATIVE AI
-

Veo 3 Video Generation Tool Format Limitations Explained
By
–
Here it is with Veo 3 integration, problem with Veo3 it's still only 16:9 (widescreen) so you can't really do portrait videos unless you crop them https://t.co/n0F59dbCUC pic.twitter.com/4VQCqyDoxX
— @levelsio (@levelsio) 10 septembre 2025Here it is with Veo 3 integration, problem with Veo3 it's still only 16:9 (widescreen) so you can't really do portrait videos unless you crop them
-
Stable Audio 2.5 Now Available on Replicate Platform
By
–
🎼Stable Audio 2.5 from @StabilityAI is live on @replicate🎶
— Replicate (@replicate) 10 septembre 2025
up to 3min tracks in just a few seconds
Studio-quality output
Commercial licensing included
Audio inpainting & extensions
Multi-part compositions
Run it via API or web interface → https://t.co/aadY5LDMvy https://t.co/na7ah1HnUFStable Audio 2.5 from @StabilityAI is live on @replicate up to 3min tracks in just a few seconds
Studio-quality output
Commercial licensing included
Audio inpainting & extensions
Multi-part compositions Run it via API or web interface → https://
replicate.com/stability-ai/s
table-audio-2.5
… -
Model Demonstrates Image Generation Through Spreadsheet Pixel Manipulation
By
–
Looks like pure model capability to me, presumably it figured out how to write out colored cells in the spreadsheet and then used one of its other binary tools to make a pixelated the image
-

32B Parameters Model Outperforms Larger Models Reasoning
By
–
Just 32B parameters – proving smaller models can be smarter, beating 10x larger models with deeper reasoning
-
K2 Think: Open-Source Reasoning Model 20x Faster on Cerebras
By
–
K2 Think is now available and runs fastest on Cerebras Inference at 2,000 TPS on Cerebras – 20x faster than GPU.
— Cerebras (@cerebras) 10 septembre 2025
Launched by @mbzuai and @G42ai, K2 Think is a world leading open-source reasoning model that beats GPT OSS 120B and DeepSeek on math and reasoning. pic.twitter.com/fsRl5QhlaXK2 Think is now available and runs fastest on Cerebras Inference at 2,000 TPS on Cerebras – 20x faster than GPU. Launched by @mbzuai and @G42ai
, K2 Think is a world leading open-source reasoning model that beats GPT OSS 120B and DeepSeek on math and reasoning. -
Generative and Agentic AI Transforming Financial Services
By
–
💳 How is Generative + Agentic AI reshaping financial services?
— NVIDIA AI (@NVIDIAAI) 10 septembre 2025
Featuring: Prem Natarajan, EVP, Chief Scientist, and Head of AI at @CapitalOne
On the NVIDIA AI Podcast, he shares how his team is building AI-powered systems to:
⚡ Reduce cognitive burden for customers +… pic.twitter.com/krKdqPM8FzHow is Generative + Agentic AI reshaping financial services? Featuring: Prem Natarajan, EVP, Chief Scientist, and Head of AI at @CapitalOne On the NVIDIA AI Podcast, he shares how his team is building AI-powered systems to: Reduce cognitive burden for customers +
-

Lucid Origin Model Now Available on Replicate Platform
By
–
Lucid Origin is now on Replicate Excellent prompt adherence and text rendering for HD output The latest model by @leonardoai
-
Stable Audio 2.5 Now Available on Multiple Platforms
By
–
You can try Stable Audio 2.5 now at StableAudio.com and the Stability AI API, as well as through our platform partners 👉 @Fal, @Replicate, and @ComfyUI.
→ View original post on X — @stabilityai, 2025-09-10 14:28 UTC
-
Stable Audio 2.5 Launch: Partnership with Amp for Sound Branding
By
–
With the launch of Stable Audio 2.5, we’re also partnering with leading sound branding agency amp, part of the Landor Group, a @WPP company, to co-develop enterprise solutions for innovative brands who want to create iconic sound identities and experiences.
→ View original post on X — @stabilityai, 2025-09-10 14:28 UTC