AI article
I put the 381 tok/s Qwen stack on one RTX 3090, real chat does 118 and the headline mode crashes
Community description: A post went around earlier this month, 82 to 381 tok/s on a single RTX 3090 running a 27B model. I...
Dev.to | Oct 16, 2026 | prodbymarcu
Automated excerpt
Twice, because the first attempt died on a BIOS setting. The baseline is the same 27B model as a Q4_K_M GGUF under llama. cpp, ~42 tok/s on the same card. Same server, same prompt, the count matches the engine's own accounting.
Selected automatically from source text; not independently written or fact-checked. Read the original for full context.