Mistral Large 4 just landed: 1 trillion parameters, trained on only ~4,000 Nvidia GPUs. That's 2-3x less than Chinese rivals, per Mistral. The pitch: a 'third way' between closed US models and open Chinese ones, aimed at security, finance and chips. Open weights are due in ~3 weeks. Want posts about news like this? www.contentbuffer.com Would you pick a European model over US or Chinese ones? #AI #Mistral #LLM #OpenSource #TechNews #ArtificialIntelligence #Europe #Startups