DeepSeek-V3 now runs at 20 tokens per second on Mac Studio, and that is a nightmare for OpenAI

Be part of our each day and weekly newsletters for the newest updates and unique content material on industry-leading AI protection. Be taught Extra Chinese language AI startup DeepSeek has quietly launched a brand new massive language mannequin that’s already sending ripples by way of the synthetic intelligence {industry} — not only for its capabilities, however for the way it’s being deployed. The 641-gigabyte mannequin, dubbed DeepSeek-V3-0324, appeared on AI repository Hugging Face at present with just about no announcement, persevering with the corporate’s sample of low-key however impactful releases. What makes this launch notably notable is the mannequin’s MIT license — making it freely obtainable for industrial use — and early reviews…