

C'EST CONFIRMÉ – GPT 5
– GPT 5 mini
– GPT 5 nano 3 versions du modèle

By
–


C'EST CONFIRMÉ – GPT 5
– GPT 5 mini
– GPT 5 nano 3 versions du modèle
By
–
Jeudi, c’est l’heure de vérité. Il y a de fortes chances que GPT-5 ne sera pas entièrement disponible en Europe. Le modèle le plus puissant sera restreint par l’AI Act de l’UE… sauf si Altman trouve une parade de contorsionniste.

By
–
Skywork UniPic Unified Autoregressive Modeling for Visual Understanding and Generation
By
–
ByteDance presents Seed Diffusion
— AK (@_akhaliq) 6 août 2025
A Large-Scale Diffusion Language Model with High-Speed Inference pic.twitter.com/32hQiBdHxZ
ByteDance presents Seed Diffusion A Large-Scale Diffusion Language Model with High-Speed Inference
By
–
Do you have a better way of coming up with a training cost estimate?
By
–
Could be the model is getting confused because it has special training about a "search" tool baked into the model https://
github.com/openai/gpt-oss
?tab=readme-ov-file#browser
… Frustrating that it doesn't notice when the search tool hasn't been configured though
By
–
North has security and data privacy at its core, designed for regulated industries and government agencies. Its lightweight architecture is optimized for private deployments with as few as two GPUs for performance on-prem or in hybrid clouds, VPCs, or airgapped environments.
By
–
Genie 3 ressemble à un moment charnière pour les world models : nous pouvons désormais générer des simulations interactives en temps réel, de plusieurs minutes, dans n’importe quel monde imaginable.
— VISION IA (@vision_ia) 6 août 2025
Cela pourrait bien être la pièce manquante pour une AGI incarnée…
Ce que… pic.twitter.com/5Nw5BBQjQn
Genie 3 ressemble à un moment charnière pour les world models : nous pouvons désormais générer des simulations interactives en temps réel, de plusieurs minutes, dans n’importe quel monde imaginable. Cela pourrait bien être la pièce manquante pour une AGI incarnée… Ce que
By
–
My guess is that the fraction of humans who understand enough about LLMs to ask those questions overlaps pretty heavily with the fraction of humans who know that LLMs are almost universally terrible at accurately answering questions about themselves
By
–
Another exciting Wan 2.2 model for you, for when speed matters most:https://t.co/DGK1nX2jRR
— Replicate (@replicate) 6 août 2025
image-to-video and text-to-video 💥
480p: 5 seconds at 24fps in 5 seconds, $0.0125 per video
720p: 5 seconds at 24fps in 12 seconds, $0.025 per video pic.twitter.com/LzT2ZWipDo
Another exciting Wan 2.2 model for you, for when speed matters most: https://
replicate.com/wan-video/wan-
2.2-5b-fast
… image-to-video and text-to-video 480p: 5 seconds at 24fps in 5 seconds, $0.0125 per video
720p: 5 seconds at 24fps in 12 seconds, $0.025 per video