Machine
680m · Ryzen 5 6600H #b420cc10
21 runs of 14 models since Oct 30, 2026. Last run Nov 14, 2026.
21
runs
14
models
14
configs
4
configs run 2+ times
Hardware
- GPU
- AMD Radeon 680M 4 GB
- CPU
- AMD Ryzen 5 6600H with Radeon Graphics 6 cores
- RAM
- 11 GB
- OS
- Arch Linux
- Tools
- llama.cpp 21
- 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, 14 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
Qwen3.6-35B-A3B-UD-Q2_K_XL
1 run · 1 config
Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL
2 runs · 1 config
★ bestQ2_Kllama.cpp8k ctxthinking off
×2 runs
75.469.4–81.4
varies
14.0 tok/s
14.0–14.1
Cyber-Tiel-Coder-35B-A3B-MTP-IQ3_S
1 run · 1 config
cyber-tiel-coder-q2kxl
1 run · 1 config
Cyber-Tiel-Coder-35B-A3B-UD-IQ3_S
5 runs · 1 config
★ bestllama.cpp8k ctxthinking off
×5 runs
71.766.7–79.3
varies
11.2 tok/s
10.7–11.6
Qwen3.6-35B-A3B-UD-IQ3_S
2 runs · 1 config
★ bestllama.cpp8k ctxthinking off
×2 runs
71.468.2–74.7
steady
11.2 tok/s
11.2–11.2
ds-v4-mtp-q4km
2 runs · 1 config
★ bestllama.cpp8k ctxthinking default
×2 runs
71.067.9–74.2
steady
8.9 tok/s
8.8–8.9
Qwen3.5-9B-Defiant-Fable-MTP-Q4_K_M
1 run · 1 config
Qwen3.6-35B-A3B-UD-IQ3_XXS
1 run · 1 config
Underdog-Saluki-27B-1.0-IQ2-mix
1 run · 1 config
Compare with another machine
Run history
Nov 14, 2026
23:01ds-v4-mtp-q4kmllama.cpp8.9 tok/s74.219:46cyber-tiel-coder-q2kxlllama.cpp12.9 tok/s72.1Nov 13, 2026
15:32ds-v4-mtp-q4kmllama.cpp8.8 tok/s67.9Oct 31, 2026
22:58Underdog-Saluki-27B-1.0-IQ2-mixllama.cpp3.5 tok/s65.319:39Cyber-Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL×2llama.cpp14.0 tok/s75.417:31Qwen3.6-35B-A3B-UD-Q2_K_XLllama.cpp13.4 tok/s76.617:21Qwen3.6-35B-A3B-UD-IQ3_XXSllama.cpp12.0 tok/s65.712:29Qwen3.6-35B-A3B-UD-IQ3_Sllama.cpp11.2 tok/s74.711:07LFM2.5-8B-A1B-Q3.8-TBrilliance-NEO-MAX-Q8_0llama.cpp15.6 tok/s51.609:34Cyber-Tiel-Coder-35B-A3B-UD-IQ3_S×3llama.cpp11.2 tok/s71.7Oct 30, 2026
23:37Cyber-Tiel-Coder-35B-A3B-UD-IQ3_S×2llama.cpp11.1 tok/s73.021:54Cyber-Tiel-Coder-35B-A3B-MTP-IQ3_Sllama.cpp11.2 tok/s73.920:49Qwen3.6-35B-A3B-UD-IQ3_Sllama.cpp11.2 tok/s68.216:53Qwen3.5-9B-Defiant-Fable-MTP-Q4_K_Mllama.cpp6.0 tok/s68.115:53LFM2.5-24B-A2B-Q4_K_Mllama.cpp19.7 tok/s57.615:41LFM2.5-24B-A2B-TBrill-NEO-IQ4_XSllama.cpp21.0 tok/s51.015:20Ternary-Bonsai-2-27B-PQ2_0llama.cpp4.0 tok/s50.1