Machine

RX 9070 9070 Xt 9070 Gre · Ryzen 7 5800X #a6c6f4d2

21 runs of 15 models since Sep 11, 2026. Last run Oct 5, 2026.

ShareRedditX
21
runs
15
models
18
configs
3
configs run 2+ times

Hardware

GPU
AMD Radeon RX 9070/9070 XT/9070 GRE 15 GB
CPU
AMD Ryzen 7 5800X 8-Core Processor 8 cores
RAM
46 GB
OS
CachyOS
Tools
LM Studio 15 · Strata 3 · llama.cpp 3
Client
0.4.59
All AMD Radeon RX 9070/9070 XT/9070 GRE 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 read4050607080901002050tok/s (log scale) →quality →

Medians of each model's best config, 15 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.

qwen3.8-flash-next-iq3_xxs
1 run · 1 config
★ bestStrata202k ctxthinking defaulteffort xhigh
89.2
1 rated run
35.4 tok/s
qwen3.8-27b-gsq-rco
2 runs · 2 configs
★ bestLM Studio94k ctxthinking default
88.8
1 rated run
31.7 tok/s
tiel-coder-35b-a3b
2 runs · 1 config
★ bestLM Studio256k ctxthinking default
×2 runs
86.085.7–86.2
steady
28.6 tok/s
28.4–28.8
occamy-1.0-Q6_K
1 run · 1 config
★ bestQ6_Kllama.cpp256k ctxthinking default
85.9
1 rated run
30.4 tok/s
qwen3.8-flash-next-iq2_xs
1 run · 1 config
★ bestStrata256k ctxthinking defaulteffort xhigh
85.7
1 rated run
45.7 tok/s
qwen3.8-flash-next-coder-iq1_m
1 run · 1 config
★ bestStrata256k ctxthinking defaulteffort xhigh
84.0
1 rated run
43.3 tok/s
dirk-qwen3.8-27b@iq3_xxs
3 runs · 2 configs
★ bestLM Studio120k ctxthinking default
×2 runs
83.582.4–84.6
steady
18.6 tok/s
18.5–18.7
Swift-Qwen3.8-27B-Uncensored-Dynamic-MTP-UD-IQ3_XXS
1 run · 1 config
★ bestllama.cpp128k ctxthinking defaulteffort xhigh
82.9
1 rated run
32.3 tok/s
tiel-coder-35b-a3b-mtp
2 runs · 1 config
★ bestLM Studio256k ctxthinking default
×2 runs
82.882.0–83.7
steady
26.3 tok/s
25.4–27.2
dirk-qwen3.8-27b
1 run · 1 config
★ bestLM Studio115k ctxthinking default
82.5
1 rated run
20.6 tok/s

Compare with another machine

Run history