Devstral-Small-2507
Performance benchmark · measured on 27.07.2026 20:34
Benchmark-ID
run-20260728-034051-4e3bc7Timebench 3 - Kombi (Prefill + Generation)Dense24BRuntime: llama.cppQuantisierung: Q4_K_M
Generation647,58tok/s
Prefill10.804,25tok/s
Time to First Token10.971,00ms
Total duration65,79s
Concurrency10parallel
