Machine

RTX 3080 Ti · Ryzen 7 5800XT #a1fe7b5e

29 runs of 22 models since Nov 22, 2026. Last run Nov 22, 2026.

ShareRedditX
29
runs
22
models
24
configs
5
configs run 2+ times

Hardware

GPU
NVIDIA GeForce RTX 3080 Ti 12 GB
CPU
AMD Ryzen 7 5800XT 8-Core Processor 8 cores
RAM
31 GB
OS
Manjaro Linux
Tools
OpenAI-compatible 26 · llama.cpp 3
Client
0.4.99
All NVIDIA GeForce RTX 3080 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 read30507090102050100tok/s (log scale) →quality →

Medians of each model's best config, 22 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.

kintsugi-27b
1 run · 1 config
★ bestOpenAI-compatible256k ctxthinking default
84.4
1 rated run
54.4 tok/s
inkling-2-mini-128k
1 run · 1 config
★ bestOpenAI-compatible128k ctxthinking default
83.0
1 rated run
72.5 tok/s
qwen3.8-27b-Swift1.5-Uncensored-ATX-AI3
2 runs · 1 config
★ bestOpenAI-compatible64k ctxthinking default
×2 runs
82.479.9–84.9
steady
91.3 tok/s
90.9–91.8
qwen3.6-35b-256k
1 run · 1 config
★ bestOpenAI-compatible256k ctxthinking default
81.9
1 rated run
36.0 tok/s
qwen3.6-35b-200k
1 run · 1 config
★ bestOpenAI-compatible200k ctxthinking default
80.7
1 rated run
35.8 tok/s
retry-qwen3.6-35b-q4-128k
1 run · 1 config
★ bestOpenAI-compatible128k ctxthinking default
80.0
1 rated run
23.6 tok/s
inkling-2-mini
1 run · 1 config
★ bestOpenAI-compatible64k ctxthinking default
79.7
1 rated run
79.3 tok/s
retry-qwen36-27b-iq3xxs
1 run · 1 config
★ bestOpenAI-compatible32k ctxthinking default
79.1
1 rated run
6.8 tok/s
qwen3.6-35b-128k
1 run · 1 config
★ bestOpenAI-compatible128k ctxthinking default
78.6
1 rated run
35.5 tok/s
deepseek-v4-pro-9b
1 run · 1 config
★ bestOpenAI-compatible32k ctxthinking default
76.3
1 rated run
112 tok/s

Compare with another machine

Run history