quinta - Stefano Quintarelli on Nostr: DeepSeek-V3 now runs at 20 tokens per second on Mac Studio, and that’s a nightmare ...
DeepSeek-V3 now runs at 20 tokens per second on Mac Studio, and that’s a nightmare for OpenAI https://blog.quintarelli.it/2025/03/deepseek-v3-now-runs-at-20-tokens-per-second-on-mac-studio-and-thats-a-nightmare-for-openai/