Côté hardware, la Chine ne joue pas petit bras. DeepSeek fait encore parler de lui. Leur V3 est un monstre de 641 Go… qui tourne sur un Mac Studio.
Open source, licence MIT, pas besoin de datacenter….
Un LLM local et massif. La Chine pousse une autre vision : puissante,
LLMS
-
DeepSeek V3: 641GB Open Source LLM Running Local Mac
By
–
-
OpenAI Adopts Anthropic’s Model Context Protocol for ChatGPT Integration
By
–
OpenAI announced it will adopt Anthropic’s open-source Model Context Protocol (MCP)
— Rowan Cheung (@rowancheung) 27 mars 2025
The move will enable ChatGPT and other OpenAI products to integrate with external data and appspic.twitter.com/AgvcpAYI9zOpenAI announced it will adopt Anthropic’s open-source Model Context Protocol (MCP) The move will enable ChatGPT and other OpenAI products to integrate with external data and apps
-
Alibaba releases Qwen2.5-Omni-7B multimodal AI model
By
–
Alibaba dropped Qwen2.5-Omni-7B, a new multimodal AI
— Rowan Cheung (@rowancheung) 27 mars 2025
—Processes across text, audio, image, and video in real-time
—Text and speech outputs
—Strong speech understanding, outperforms specialized audio models
—Can run on phones, laptops
—Open-sourcepic.twitter.com/WpOTdqv0N6Alibaba dropped Qwen2.5-Omni-7B, a new multimodal AI —Processes across text, audio, image, and video in real-time
—Text and speech outputs
—Strong speech understanding, outperforms specialized audio models
—Can run on phones, laptops
—Open-source -
ChatGPT 4o posts screenshots of its own outputs
By
–
ChatGPT 4o generates a screenshot of it posting the screenshot it’s generating to X dot com, where it is poorly received
-
Claude Sonnet 3.7 MAX on Cursor 500k? Enterprise Only?
By
–
It is yet unclear if this will be offered to Enterprise customers only, as some reports suggest it is already the case. Is Claude Sonnet 3.7 MAX on Cursor 500k? Full scoop
-
Anthropic to Increase Claude Sonnet 3.7 Context Window to 500K
By
–
BREAKING : Anthropic is preparing to release a new version of Claude Sonnet 3.7 with a 500K Context Window (Current is 200K).
-
ChatGPT desktop vs IDE and MCP: partial competition
By
–
I think it is both yes and no – ChatGPT desktop app is unlikely will meant to be an IDE, but it is meant to take over some use cases. As for MCP – sort of yes – they will compete with other MCP operators (similar example Claude desktop app)
-
Debunking AI Myths: LLMs Continue Advancing Beyond Steady State
By
–
Got asked recently what the #1 AI myth I want to debunk is. It's this idea of the “steady state" 3 years ago: LLMs will always be toxic/biased
2 years ago: LLMs will never be empathetic
1 year ago (everyday tbh): LLMs are running out of gains Steady state is far behind us -
Pointwise vs Cumulative Metrics in Language Model Evaluation
By
–
Good catch! The numbers in the old table were pointwise estimates – pointwise performance is a bucketed estimate over context lengths, while the paper reports a cumulative average over context lengths. Pointwise and cumulative metrics are naturally incomparable and the pointwise
