Machine

RX 7900 XTX · Ryzen Threadripper 3970X #213b1ee3

26 runs of 11 models since Nov 3, 2026. Last run Nov 8, 2026.

ShareRedditX
26
runs
11
models
23
configs
3
configs run 2+ times

Hardware

GPU
AMD Radeon RX 7900 XTX 24 GB
CPU
AMD Ryzen Threadripper 3970X 32-Core Processor 32 cores
RAM
94 GB
OS
Ubuntu 26.04.1 LTS
Tools
OpenAI-compatible 19 · Strata 7
Client
0.4.99
All AMD Radeon RX 7900 XTX 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, 11 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-Avalon-IQ3KS-00001-of-00033
1 run · 1 config
★ bestOpenAI-compatible64k ctxthinking default
88.3
1 rated run
61.7 tok/s
strata_iq3s_128k
1 run · 1 config
★ bestOpenAI-compatible64k ctxthinking on
87.2
1 rated run
53.7 tok/s
strata_iq3s_256k
2 runs · 2 configs
★ bestOpenAI-compatible32k ctxthinking on
86.8
1 rated run
39.8 tok/s
strata_ud-q5
1 run · 1 config
★ bestOpenAI-compatible64k ctxthinking on
86.4
1 rated run
43.0 tok/s
qwen3.8-flash-next-unsloth-ud-iq4_xs-v31
6 runs · 6 configs
★ bestStrata256k ctxthinking defaulteffort xhigh
86.3
1 rated run
64.6 tok/s
default
4 runs · 4 configs
★ bestOpenAI-compatible256k ctxthinking on
86.3
1 rated run
68.1 tok/s
strata_swift
4 runs · 3 configs
★ bestOpenAI-compatible32k ctxthinking on
×2 runs
84.684.4–84.8
steady
56.3 tok/s
55.6–57.0
ThinkingCap-Qwen3.8-27B
2 runs · 2 configs
★ bestOpenAI-compatible32k ctxthinking on
84.4
1 rated run
46.5 tok/s
strata_iq4
2 runs · 1 config
★ bestOpenAI-compatible32k ctxthinking on
×2 runs
83.880.5–87.2
steady
60.1 tok/s
54.3–66.0
cyber-frost-3.8-ps
2 runs · 1 config
★ bestStrata64k ctxthinking defaulteffort xhigh
×2 runs
83.783.4–84.1
steady
27.0 tok/s
24.1–29.9

Compare with another machine

Run history