GPT-oss models running under Vulkan: Benchmark Framework Desktop Mainboard and 4-node cluster · Issue #21 · geerlingguy/ollama-benchmark · GitHub
20b: 45 t/s single node - 97W
120b: 33 t/s single node - 98W
120b: 24 t/s clustered (4x) - 138W
11 Likes