Skip to content
llm-speed
← raw run
wrapped·Oct 22, 2026
/r/r_klxxsjjp2m-
your run

Flint-40B-A2.5B-UD-cpumoe6-ctx8k

on RTX 3090
16.8
tok/sdecode
RTX 3090 (24GB) + AMD EPYC 7763 64-Core Processor (128c) + 1008GB
rank in tier
21/21RTX 3090 runstop 100%
best workload
chat-short
where the rig flew
slowest workload
concurrent-decode16.8 tok/s
where the rig struggled
backend
llama.cppllama.cpp is the universal backend — broad model support, modest speed ceiling.
faster than
  • llama-3.3-70b-Instruct-4bit on M3 Ultra16.8 tok/s
  • qwen2.5-72b-Instruct-4bit on M3 Ultra16.3 tok/s
  • phi-4-4bit on M3 Pro15.6 tok/s
share
16.8 tok/s on RTX 3090 running Flint-40B-A2.5B-UD-cpumoe6-ctx8k — top 100% of RTX 3090 runs. Signed benchmark via llm-speed.
https://llm-speed.com/wrapped/r_klxxsjjp2m-
X / TwitterBlueskyMastodonRedditraw run

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.