Skip to content
llm-speed

qwen2.5-coder on RTX 3090 (24GB) + AMD EPYC 7702P 64-Core Processor (64c) + 252GB

Workload results

WorkloadBackendModeldecode tok/sprefill tok/sTTFTp50p95
chat-shortollama@0.31.1qwen2.5-coderQ4_K_M69.21tok/s402.0tok/s326ms14.4ms14.9ms
chat-longollama@0.31.1qwen2.5-coderQ4_K_M65.83tok/s2,070.6tok/s1,530ms15.2ms15.6ms
concurrent-decodeollama@0.31.1qwen2.5-coderQ4_K_M67.64tok/s14.7ms15.3ms
agent-traceollama@0.31.1qwen2.5-coderQ4_K_M63.63tok/s3,518.8tok/s483ms15.4ms17.4ms

Reproduce on your machine

Same workload, same model, signed at your rig. The exact command that produced this run:

$ pipx install llm-speed && llm-speed bench --model 'qwen2.5-coder' --workload 'chat-short'

Runs in about a minute. Your number lands on the leaderboard signed and linkable. How it's measured.

Embed this run

Drop the badge into a README, blog post, or signature. Each render is a backlink to the signed result.

llm-speed: 69.2 tok/s on RTX 3090 (24GB) (qwen2.5-coder)
[![llm-speed: 69.2 tok/s on RTX 3090 (24GB) (qwen2.5-coder)](https://llm-speed.com/badge/r_6ahy-dq0f_0.svg)](https://llm-speed.com/r/r_6ahy-dq0f_0)

Related benchmarks

Provenance

Run ID
r_6ahy-dq0f_0
Fingerprint hash
Public key
+NwKPaMRjmIfrqAm7KKOOld2RbM1lTa0cHfi0ARmmlo=
Received
2026-07-02 05:12:05