Skip to content
llm-speed

llm-speed

The benchmark suite for local + hosted LLM inference.Run it. See your tok/s. Compare your rig.

$ pipx install llm-speed && llm-speed bench

Runs in 60 seconds. Auto-detects your hardware. Open source. · Pre-launch: pipx install https://api-dev.llm-speed.com/dist/llm_speed-0.0.1-py3-none-any.whl

Methodology →Privacy →About →

Latest submissions

13 runs
  • r_a3rufd3qvwi
    signed
    M3 Ultra (60-core GPU) + 96GB unified
    suite-v1
  • r_jy6em3a4pck
    signed
    M3 Ultra (60-core GPU) + 96GB unified
    suite-v1
  • r_rb3siysn95e
    signed
    M3 Pro (18-core GPU) + 36GB unified
    suite-v1
  • r_4gkiuiaapay
    signed
    M3 Pro (18-core GPU) + 36GB unified
    suite-v1
  • r_wmulvjary_u
    signed
    M3 Ultra (60-core GPU) + 96GB unified
    suite-v1
  • r_eeaswwp6sp4
    signed
    M3 Ultra (60-core GPU) + 96GB unified
    suite-v1
  • r_ie1wp1y1thw
    signed
    M3 Ultra (60-core GPU) + 96GB unified
    suite-v1
  • r_qnz2zuxa3jo
    signed
    M3 Pro (18-core GPU) + 36GB unified
    suite-v1
  • r_e0s_1vwghwu
    signed
    M3 Ultra (60-core GPU) + 96GB unified
    suite-v1
  • r_zzlpyzzkgg8
    signed
    M3 Ultra (60-core GPU) + 96GB unified
    suite-v1
  • r_buygyuyludy
    signed
    M3 Pro (18-core GPU) + 36GB unified
    suite-v1
  • r_-dmetqtv_he
    signed
    M3 Pro (18-core GPU) + 36GB unified
    suite-v1
  • r_iqtwf20dfmo
    signed
    M3 Pro (18-core GPU) + 36GB unified
    suite-v1