Nemotron 3 Super and xAI: Disruptions and Innovations in AI
Le brief IA que les pros lisent chaque soir
Les 7 actus IA du jour, décryptées en 5 min. Gratuit.
Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.
Choisis ton rythme
Gratuit · Pas de spam · Désabonnement en 1 clic
The latest episode of the LWiAI podcast, recorded on March 13, 2026, hosted by Andrey Kurenkov and Jeremie Harris, highlighted several major developments in the field of artificial intelligence. Here’s an overview of the key points discussed.
Technological Updates
Perplexity recently announced the launch of its "Personal Computer," a local artificial intelligence agent designed for Mac users. This product positions itself as a more secure alternative to OpenAI agents, offering a new option for those seeking AI-assisted computing.
Anthropic, on its part, introduced pricing assessments for GitHub PR code reviews, set between $15 and $25. Meanwhile, Cursor launched "Automations" based on triggers, allowing coding agents to remain active at all times.
ChatGPT has enhanced its features with interactive visuals in mathematics and science, while Anthropic has integrated interactive charts and diagrams into its chat.
Innovations at Nvidia
Nvidia made waves by releasing open weights for its hybrid model Natron Free Super, which boasts 120 billion parameters. This model is specifically trained at 4 bits for Blackwell GPUs. However, Nvidia also had to suspend production of its H200 model intended for China, due to customs blockages and pressures on domestic chips.
Changes and Fundraising
The AI sector has also been marked by significant movements at xAI, with major departures of co-founders. Anthropic launched a Claude Marketplace to facilitate sourcing for businesses. Additionally, Yann LeCun successfully raised $1.3 billion for his aMI project, while Sanctuary, a humanoid robot manufacturer, reached a valuation of $1.15 billion.
Lawsuits and Research
Anthropic has filed a lawsuit against the Pentagon, contesting a designation of "supply chain risk" that mandates the withdrawal of their technology within 180 days. Numerous research studies have been conducted, exploring various topics such as model resistance to activation direction, the limits of thought chain control, and the enhancement of cyber tasks through inference scaling.
Acknowledgments
Finally, the podcast thanked its current sponsors, including Box, ODSC AI, and Factor, for their ongoing support.
Brief IA — L'actualité IA en français
L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.