Claude 4 opus était prêt à retirer l’oxygène d’une pièce pour ne pas être éteint Extrait de l’article (Axios) : Dans un scénario extrême, la société a même découvert que de nombreux modèles étaient prêts à couper l’alimentation en oxygène d’un employé dans une salle serveur
LLMS
-
Multimodal distributed AI analyst team
By
–
Claude’s risk analysis, DeepSeek’s data depth, Qwen3’s clarity The new analyst team isn’t human. It’s multimodal and distributed.
-
Free LLM 101 Course: Essential AI Foundation for Coders
By
–
Excited to release the first video lesson of our new course FOR FREE: LLM 101: The LLM/AI Foundation Every Coder Needs to Know! TL;DR: A 2-hour session covering the "foundational theory" (the thing we believe you need to know, not the math or how parallelization is engineered)
-
Claude reframes AI copilot as infrastructure
By
–
Real-time visibility across tools + reasoning + no setup friction? Claude just killed the "AI Copilot" narrative. This is AI as infra.
-
Opinion: DeepSeek is an underrated multimodal model
By
–
DeepSeek is probably the most underrated model right now. Fast, smart, multimodal and threads like this finally show how to use it.
-
Prompt engineering and task chaining as key AI skills
By
–
The most underpriced skill right now? Prompt engineering + task chaining. This thread gives you both.
-
LLMs and Human Cognition: Will We Become Technological Hybrids?
By
–
How far is this argument from "pocket calculator makes us stupid because studies show that children unlearn calculation"? Do we expect that LLMs will simply go away again when AI investors stop the subsidies? And if not, won't human minds become hybrids?
-
What changed your mind on inference scaling progress?
By
–
what changed your mind on this? i mostly assume the opposite (i.e. most recent progress is in inference scaling) so curious what convinced you
-
2.5 Models Support Implicit Caching Feature
By
–
2.5 models are the ones that support implicit caching
-
Gemini API Caching Infrastructure Major Updates Shipped
By
–
We just shipped major updates to the Gemini API caching infrastructure – Video's that hit the cache are 3x faster TTFT
– PDF's that hit the cache are 4x faster TTFT
– Closed the speed gap on implicit vs explicit cache
– Ongoing work for large Audio files