Machine

RTX 3070 Ti · Ryzen 9 9900X3D #f6347364

34 runs of 14 models since Oct 2, 2026. Last run Nov 1, 2026.

ShareRedditX
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
All NVIDIA GeForce RTX 3070 Ti benchmarks

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
below 20 tok/s: slower than you read4050607080901002050100tok/s (log scale) →quality →outliers sit at the edges

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.

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
★ bestUD-Q4_K_XLUnsloth Studio32k ctxthinking default
76.7
1 rated run
60.9 tok/s
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
★ bestUD-Q4_K_XLUnsloth Studio8k ctxthinking default
66.3
1 rated run
57.7 tok/s
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
★ bestIQ4_XSUnsloth Studio8k ctxthinking default
62.2
1 rated run
104 tok/s
unsloth/Qwen3.5-4B-MTP-GGUF
1 run · 1 config
★ bestQ8_0Unsloth Studio8k ctxthinking default
61.2
1 rated run
15.9 tok/s

Compare with another machine

Run history

Show all 34 runs ▾