Machine
Ryzen 5 5600G #561b1b59
37 runs of 7 models since Nov 13, 2026. Last run Nov 21, 2026.
37
runs
7
models
13
configs
8
configs run 2+ times
Hardware
- CPU
- AMD Ryzen 5 5600G with Radeon Graphics 6 cores
- RAM
- 31 GB
- OS
- Ubuntu 24.04.5 LTS
- Tools
- llama.cpp 37
- Client
- 0.4.99
Models on this machine
One dot per model, at its best config. Up is better answers, right is faster. Time to finish counts every token the model wrote, thinking included, so a model that talks a lot ends up further left.
Best trade-off: quality 60+, and nothing else here is both faster and betterOther models · dot size = number of runs
Medians of each model's best config, 7 models.
Best config per model
sort
Runs with the same model, quant, tool, context, thinking and KV cache count as one config. Numbers are the median of its runs, and each dot is one run on the 0–100 quality scale. ★ marks the best config with at least 2 rated runs.
configrunsqualityeach runspeed
qwen38-27b-ISTA-DASLab-GSQ-RCO-IQ3_s
1 run · 1 config
Qwen3.8-Flash-Next-Avalon-IQ3KS-00001-of-00033
22 runs · 5 configs
★ bestllama.cpp32k ctxthinking oneffort medium
×10 runs
84.481.2–88.5
steady
42.8 tok/s
36.3–45.0
Qwen3.6-35B-A3B-UD-Q3_K_S
1 run · 1 config
Inkling-2-Mini-UD-IQ3_XXS
3 runs · 1 config
★ bestllama.cpp32k ctxthinking on
×3 runs
80.266.0–84.4
varies
42.8 tok/s
34.4–43.4
Qwen3.6-35B-A3B-UD-IQ2_M
2 runs · 1 config
★ bestllama.cpp96k ctxthinking default
×2 runs
78.677.0–80.2
steady
110 tok/s
107–114
Qwen3.6-35B-A3B-UD-IQ3_XXS
1 run · 1 config
qwen36-agent
7 runs · 3 configs
★ bestllama.cpp96k ctxthinking on
×3 runs
56.653.2–75.9
1 outlier
96.6 tok/s
96.2–97.5
One run of the best config scored 75.9, the other 2 sit between 53.2 and 56.6. The median isn't thrown off by it.
Compare with another machine
Run history
Nov 20, 2026
18:18Qwen3.8-Flash-Next-Avalon-IQ3KS-00001-of-00033×10llama.cpp42.8 tok/s84.414:10Qwen3.8-Flash-Next-Avalon-IQ3KS-00001-of-00033llama.cpp41.7 tok/s82.713:30Qwen3.8-Flash-Next-Avalon-IQ3KS-00001-of-00033llama.cpp42.7 tok/s82.311:24qwen36-agentllama.cpp97.5 tok/s75.9Nov 19, 2026
21:41qwen36-agent×2llama.cpp96.4 tok/s54.921:30qwen36-agentllama.cpp96.7 tok/s61.821:21qwen36-agent×3llama.cpp96.8 tok/s56.5Nov 17, 2026
22:41Qwen3.6-35B-A3B-UD-Q3_K_Sllama.cpp59.7 tok/s81.621:18Qwen3.6-35B-A3B-UD-IQ3_XXSllama.cpp100 tok/s73.721:02Qwen3.6-35B-A3B-UD-IQ2_M×2llama.cpp110 tok/s78.6Show all 37 runs ▾Show fewer ▴
Nov 14, 2026
06:30Qwen3.8-Flash-Next-Avalon-IQ3KS-00001-of-00033llama.cpp41.2 tok/s44.406:04Qwen3.8-Flash-Next-Avalon-IQ3KS-00001-of-00033llama.cpp43.7 tok/s82.7Nov 13, 2026
22:52Qwen3.8-Flash-Next-Avalon-IQ3KS-00001-of-00033×2llama.cpp36.9 tok/s84.822:33Qwen3.8-Flash-Next-Avalon-IQ3KS-00001-of-00033llama.cpp38.2 tok/s79.722:22Qwen3.8-Flash-Next-Avalon-IQ3KS-00001-of-00033llama.cpp40.8 tok/s84.122:11Qwen3.8-Flash-Next-Avalon-IQ3KS-00001-of-00033llama.cpp40.1 tok/s88.521:35Qwen3.8-Flash-Next-Avalon-IQ3KS-00001-of-00033llama.cpp37.5 tok/s87.921:25Qwen3.8-Flash-Next-Avalon-IQ3KS-00001-of-00033llama.cpp44.4 tok/s64.5