Skip to content
llm-speed
← raw run
wrapped·Jul 02, 2026
/r/r_o2-1w665rtq
your run

coder-v2

on RTX 3090
189.5
tok/sdecode
RTX 3090 (24GB) + AMD EPYC 7702P 64-Core Processor (64c) + 252GB
rank in tier
3/14RTX 3090 runstop 21%
best workload
chat-short
where the rig flew
slowest workload
chat-long77.6 tok/s
where the rig struggled
backend
ollamaollama is llama.cpp under the hood; pure llama.cpp tends to nudge a few % faster.
faster than
  • Qwen3-4B-Instruct-2507-4bit on M3 Ultra182.7 tok/s
  • qwen3-coder on RTX 4090179.9 tok/s
  • DeepSeek-Coder-V2-Lite-Instruct-4bit on M3 Ultra168.3 tok/s

This is a shareable view of one signed llm-speed suite-suite-v1 run. Numbers are not edited. The raw record has every workload, the public key, and the fingerprint hash.