On VoiceBench, it scores 56.8 overall with just 1.5B parameters.
MULTIMODAL AI
-
LFM2-1.2B Extension: Novel Audio Foundation Model Architecture
By
–
It has a super interesting architecture that extends LFM2-1.2B.
— Maxime Labonne @ ICLR (@maximelabonne) 1 octobre 2025
Unlike other audio foundation models, we disentangle the representations of audio input and output.
More info in the blog post. pic.twitter.com/cei9MSoCRuIt has a super interesting architecture that extends LFM2-1.2B. Unlike other audio foundation models, we disentangle the representations of audio input and output. More info in the blog post.
-
LFM2-Audio: 1.5B Multimodal Model 10x Faster Inference
By
–
LFM2-Audio just dropped!
— Maxime Labonne @ ICLR (@maximelabonne) 1 octobre 2025
It's a 1.5B model that understands and generates both text and audio
Inference 10x faster + quality on par with models 10x larger
Available today on @huggingface and our playground 🥳 pic.twitter.com/iJXWpyBCgqLFM2-Audio just dropped! It's a 1.5B model that understands and generates both text and audio Inference 10x faster + quality on par with models 10x larger Available today on @huggingface and our playground
-
OpenAI Sora 2, AI Scientists, and Latest AI Tools News
By
–
Top stories in AI today: – OpenAI’s Sora 2 with social app
– Periodic Labs’ AI scientist for physical world
– Build AI productivity tools without coding
– Amazon’s Alexa+ integrated devices – 4 new AI tools, community workflows, and more Read more: https://
therundown.ai/p/sora-2-break
s-the-internet
… -
Claude 4.5 Sonnet vs ChatGPT 5: Shocking Results
By
–
I tested Claude 4.5 Sonnet and ChatGPT 5 with same critical prompts. The results will blow your mind. Claude 4.5 Sonnet Vs. ChatGPT 5 (Video demos are included)
-
Sora’s Success: Personalized AI Content Drives Adoption
By
–
I said in my video that no one wants a social feed of AI generated slop (referring to Meta Vibes). Turns out, if you allow people to create AI generated slop of themselves and their friends, that changes things quite a bit. Sora is currently the 5th most downloaded app.
-
OpenAI Announces Sora 2, Rivals Google Veo3 in Realism
By
–
ON Y EST ! Open AI annonce SORA 2 :
— VISION IA (@vision_ia) 1 octobre 2025
Le réalisme de Sora est très convaincant. OpenAI a en réalité réussi à rattraper le Veo3 de Google. Néanmoins, OpenAI va devoir supporter des coûts immenses avec Sora2. pic.twitter.com/6Q7NK9d34EON Y EST ! Open AI annonce SORA 2 : Le réalisme de Sora est très convaincant. OpenAI a en réalité réussi à rattraper le Veo3 de Google. Néanmoins, OpenAI va devoir supporter des coûts immenses avec Sora2.
-
Achieving Sub-5cm Accuracy in 3D Imaging and Wireframe Models
By
–
How is it possible to get to below 5cm accuracy as in this image? Or do the wireframe represent something else?
-
BAIR Researchers Win Best Student Paper at CoRL2025
By
–
Congratulations to BAIR Researchers from @trevordarrell @pabbeel @JitendraMalikCV @akanazawa labs who won the Best Student Paper award at the just concluded #CoRL2025 in Seoul, Korea for their paper "Visual Imitation Enables Contextual Humanoid Control." @arthurallshire
-
Wan2.2-T2V-A14B-4steps-lora-250928 Model Release
By
–
Wan2.2-T2V-A14B-4steps-lora-250928 pic.twitter.com/5vZJqNlEFQ
— AK (@_akhaliq) 30 septembre 2025Wan2.2-T2V-A14B-4steps-lora-250928