Skip to content
llm-speed
← raw run
wrapped·Oct 22, 2026
/r/r_-yeg4qnhjp7
your run

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

on RTX 3090
130.8
tok/sdecode
RTX 3090 (24GB)
rank in tier
7/21RTX 3090 runstop 33%
best workload
chat-short
where the rig flew
slowest workload
agent-trace124.7 tok/s
where the rig struggled
backend
llama.cppllama.cpp is the universal backend — broad model support, modest speed ceiling.
faster than
  • Llama-3.1-8B-Instruct-4bit on M3 Ultra130.2 tok/s
  • Qwen3-8B-4bit on M3 Ultra120.5 tok/s
  • DeepSeek-R1-0528-Qwen3-8B-MLX-4bit on M3 Ultra116.8 tok/s
share
130.8 tok/s on RTX 3090 running Flint-40B-A2.5B-UD-ctx8k — top 33% of RTX 3090 runs. Signed benchmark via llm-speed.
https://llm-speed.com/wrapped/r_-yeg4qnhjp7
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.