Skip to content
llm-speed
← raw run
wrapped·Oct 14, 2026
/r/r_t-a9v_ipdit
your run

gpt-oss-20b

on RTX 4090
242.6
tok/sdecode
RTX 4090 (24GB) + AMD EPYC 7642 48-Core Processor (48c) + 252GB
rank in tier
1/31RTX 4090 runstop 3%
best workload
chat-short
where the rig flew
slowest workload
chat-long231.2 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 on RTX 5090232.2 tok/s
  • Qwen3.6-35B-A3B-Q4_K_M.gguf on RTX 5090224.0 tok/s
  • Yi-Coder-9B-Chat on RTX 5090199.3 tok/s
share
242.6 tok/s on RTX 4090 running gpt-oss-20b — top 3% of RTX 4090 runs. Signed benchmark via llm-speed.
https://llm-speed.com/wrapped/r_t-a9v_ipdit
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.