Machine
RTX 3070 Ti · Ryzen 9 9900X3D #f6347364
34 runs of 14 models since Oct 2, 2026. Last run Nov 1, 2026.
34
runs
14
models
18
configs
6
configs run 2+ times
Hardware
- GPU
- NVIDIA GeForce RTX 3070 Ti 8 GB
- CPU
- AMD Ryzen 9 9900X3D 12-Core Processor 12 cores
- RAM
- 60 GB
- OS
- Linux Mint 22.3
- Tools
- Unsloth Studio 34
- Client
- 0.4.98
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, 13 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
unsloth/Qwen3.6-35B-A3B-MTP-GGUF
3 runs · 1 config
★ bestUD-Q4_K_MUnsloth Studio8k ctxthinking default
×3 runs
78.073.6–81.3
steady
52.3 tok/s
51.4–53.0
unsloth/LFM2.5-24B-A2B-GGUF
2 runs · 2 configs
openbmb/MiniCPM-V-4.7-35B-A3B-gguf
4 runs · 2 configs
★ bestQ4_K_MUnsloth Studio8k ctxthinking default
×3 runs
74.268.4–76.3
steady
57.2 tok/s
57.0–59.0
huihui-ai/Huihui-Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated-MTP-GGUF
4 runs · 1 config
★ bestQ5_KUnsloth Studio8k ctxthinking default
×4 runs
73.057.7–79.1
1 outlier
51.6 tok/s
50.3–53.1
One run of the best config scored 57.7, the other 3 sit between 71.6 and 79.1. The median isn't thrown off by it.
Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF
4 runs · 1 config
★ bestQ4_KUnsloth Studio8k ctxthinking default
×4 runs
70.264.9–80.4
varies
58.4 tok/s
55.1–58.9
huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF
7 runs · 3 configs
★ bestQ4_KUnsloth Studio8k ctxthinking default
×5 runs
66.360.4–77.6
varies
59.0 tok/s
56.7–59.2
unsloth/gemma-4-26B-A4B-it-qat-GGUF
1 run · 1 config
unsloth/gemma-4-E4B-it-qat-GGUF
3 runs · 1 config
★ bestUD-Q4_K_XLUnsloth Studio8k ctxthinking default
×3 runs
65.164.0–71.8
steady1 slow run
203 tok/s
120–205
HauhauCS/Gemma-4-E4B-Uncensored-HauhauCS-Aggressive
1 run · 1 config
unsloth/Qwen3.5-4B-MTP-GGUF
1 run · 1 config
Compare with another machine
Run history
Nov 1, 2026
08:32Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUFUnsloth Studio67.6 tok/s80.508:06Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUFUnsloth Studio50.7 tok/s77.9Oct 14, 2026
12:45Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUFUnsloth Studio55.1 tok/s72.212:30Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF×4Unsloth Studio58.3 tok/s67.810:06LFM2.5-24B-A2B-GGUFUnsloth Studio60.9 tok/s76.708:58LFM2.5-24B-A2B-GGUFUnsloth Studio61.5 tok/s62.508:14Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUFUnsloth Studio58.1 tok/s64.907:52MiniCPM-V-4.7-35B-A3B-gguf×3Unsloth Studio57.2 tok/s74.206:54MiniCPM-V-4.7-35B-A3B-ggufUnsloth Studio34.0 tok/s76.106:26Qwen3.6-35B-A3B-MTP-GGUFUnsloth Studio51.4 tok/s81.3Oct 4, 2026
17:36Qwen3.6-35B-A3B-MTP-GGUFUnsloth Studio52.3 tok/s73.617:24Huihui-Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated-MTP-GGUF×3Unsloth Studio52.5 tok/s71.609:43gemma-4-E4B-it-qat-GGUF×2Unsloth Studio204 tok/s68.409:31Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUFUnsloth Studio59.2 tok/s64.609:20Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUFUnsloth Studio58.7 tok/s80.409:09Huihui-Qwen3.6-35B-A3B-Claude-4.7-Opus-abliterated-MTP-GGUFUnsloth Studio50.8 tok/s74.307:54Nex-N2.5-mini-APEX-I-MiniPlus-V2.1-GGUFUnsloth Studio56.7 tok/s58.507:38Gemma-4-E4B-Uncensored-HauhauCS-AggressiveUnsloth Studio104 tok/s62.206:52Qwen3.8_4B_Distilled_GGUFUnsloth Studio28.8 tok/s49.5Oct 3, 2026
18:36GLM-4.7-Flash-GGUFUnsloth Studio29.4 tok/s60.017:48Qwen3.6-35B-A3B-MTP-GGUFUnsloth Studio53.0 tok/s78.017:22NVIDIA-Nemotron-3-Nano-4B-GGUFUnsloth Studio139 tok/s49.716:43gemma-4-26B-A4B-it-qat-GGUFUnsloth Studio57.7 tok/s66.316:20gemma-4-E4B-it-qat-GGUFUnsloth Studio120 tok/s64.0