Machine
RTX 3080 + RTX 3080 Ti · Ryzen 7 5800X #e42f80d7
35 runs of 21 models since Aug 19, 2026. Last run Nov 7, 2026.
35
runs
21
models
30
configs
3
configs run 2+ times
Hardware
- GPU
- NVIDIA GeForce RTX 3080 + NVIDIA GeForce RTX 3080 Ti 32 GB
- CPU
- AMD Ryzen 7 5800X 8-Core Processor 8 cores
- RAM
- 62 GB
- OS
- Arch Linux
- Tools
- llama.cpp 28 · SGLang 5 · vLLM 2
- 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, 21 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
cyankiwi/Qwen3.8-27B-AWQ-INT4
2 runs · 2 configs
unsloth/Qwen3.8-27B-GGUF
1 run · 1 config
unsloth/Muse-Glimmer-30B-GGUF
3 runs · 3 configs
Qwen3.8-27B-SC_6.00bpw_H6
1 run · 1 config
Jackrong/Qwopus3.8-27B-Coder-Flash-GGUF
1 run · 1 config
unsloth/Qwen3.8-Flash-Next-GGUF
1 run · 1 config
unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF
5 runs · 2 configs
★ bestUD-Q4_K_XLllama.cpp64k ctxthinking on
×4 runs
79.176.2–85.6
varies
179 tok/s
159–184
cyankiwi/gemma-4-31B-it-AWQ-4bit
4 runs · 4 configs
unsloth/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-GGUF
1 run · 1 config
./gemma-4-31B-it-AWQ4-styletune-head
1 run · 1 config
Compare with another machine
Run history
Oct 30, 2026
20:15Gemma-4-31B-StyleTune:MY-Q6_K_XLllama.cpp24.6 tok/s73.620:06Gemma-4-31B-StyleTune:MY-Q6_K_XLllama.cpp23.9 tok/s75.817:48Gemma-4-31B-StyleTune:MY-Q4_K_XLllama.cpp43.4 tok/s78.4Oct 26, 2026
20:42gemma-4-31B-it-qat-GGUF:UD-Q4_K_XLllama.cpp47.8 tok/s65.320:26unsloth--gemma-4-26B-A4B-it-UD-Q8_K_XLllama.cpp104 tok/s73.7Show all 35 runs ▾Show fewer ▴
Oct 25, 2026
08:33flint-GGUF:UD-Q4_K_XLllama.cpp132 tok/s75.0Oct 23, 2026
20:46Qwen3.8-27B-AWQ-INT4SGLang59.3 tok/s63.218:17Qwen3.8-27B-AWQ-INT4SGLang37.2 tok/s89.6Oct 22, 2026
21:06gemma-4-31B-it-AWQ-4bitSGLang32.0 tok/s78.8Oct 21, 2026
22:20gemma-4-31B-it-AWQ-4bitSGLang31.9 tok/s64.6Oct 18, 2026
20:09gemma-4-31B-it-AWQ-4bitvLLM32.7 tok/s71.5Oct 17, 2026
09:19gemma-4-31B-it-AWQ-4bitvLLM32.7 tok/s78.0Aug 29, 2026
12:38Qwen3.8-27B-SC_6.00bpw_H6llama.cpp41.8 tok/s86.409:59Qwen3.6-35B-A3B-MTP-GGUF:UD-Q8_K_XLllama.cpp67.2 tok/s73.0Aug 28, 2026
21:19gemma-4-31B-it-4.00bpwllama.cpp53.8 tok/s55.317:23Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XLllama.cpp21.5 tok/s85.7Aug 21, 2026
17:17NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:MXFP4_MOEllama.cpp162 tok/s76.514:49NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:UD-Q4_K_XLllama.cpp159 tok/s77.414:36Muse-Glimmer-30B-GGUF:UD-Q8_K_XLllama.cpp40.0 tok/s85.0