Machine
RX 7900 XTX · Ryzen 7 7800X3D #ef96b457
32 runs of 10 models since Apr 16, 2026. Last run Apr 17, 2026.
32
runs
10
models
22
configs
7
configs run 2+ times
Hardware
- GPU
- AMD Radeon RX 7900 XTX 23 GB
- CPU
- AMD Ryzen 7 7800X3D 8-Core Processor 8 cores
- RAM
- 31 GB
- OS
- Windows
- Tools
- Ollama 28 · LM Studio 4
- Client
- 0.4.15
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, 9 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
gemma4:31b
2 runs · 2 configs
gpt-oss:20b
8 runs · 4 configs
★ bestMXFP4Ollama8k ctxthinking default
×3 runs
75.475.0–77.9
steady
63.5 tok/s
57.0–69.1
qwen/qwen3.5-35b-a3b
1 run · 1 config
qwen3-coder
2 runs · 2 configs
ministral-3:14b
3 runs · 1 config
★ bestQ4_K_MOllama64k ctxthinking default
×3 runs
68.466.3–73.0
steady
41.6 tok/s
41.6–43.0
glm-4.7-flash-unsloth:small-hotfixed-90kHEAT
7 runs · 5 configs
★ bestQ3_K_SOllama16k ctxthinking default
×2 runs
67.765.2–70.1
steady
31.6 tok/s
29.4–33.7
openai/gpt-oss-20b
3 runs · 2 configs
★ bestLM Studio64k ctxthinking default
×2 runs
65.063.8–66.2
steady
109 tok/s
106–111
qwen3.5:9b
2 runs · 1 config
★ bestQ4_K_MOllama16k ctxthinking default
×2 runs
64.060.2–67.9
steady
15.1 tok/s
13.6–16.6
qwen3:14b
2 runs · 2 configs
gemma4:26b
2 runs · 2 configs
Compare with another machine
Run history
Apr 17, 2026
19:27gpt-oss-20b×2LM Studio109 tok/s65.019:21gpt-oss-20bLM Studio110 tok/s64.219:06qwen3.5:9bOllama13.6 tok/s60.218:55qwen3.5-35b-a3bLM Studio29.2 tok/s71.718:43qwen3:14bOllama13.1 tok/s61.518:16qwen3:14bOllama11.6 tok/s50.018:11gpt-oss:20bOllama57.0 tok/s75.015:57qwen3-coder:latestOllama33.1 tok/s57.515:49ministral-3:14bOllama41.6 tok/s73.015:46gpt-oss:20bOllama51.3 tok/s73.815:44qwen3-coder:latestOllama96.0 tok/s71.215:17glm-4.7-flash-unsloth:small-hotfixed-90kHEATOllama33.7 tok/s70.114:26qwen3.5:9bOllama16.6 tok/s67.914:07gpt-oss:20bOllama69.1 tok/s75.413:12ministral-3:14bOllama43.0 tok/s66.312:25glm-4.7-flash-unsloth:small-hotfixed-90kHEATOllama35.1 tok/s60.611:32gpt-oss:20bOllama60.5 tok/s73.711:07gemma4:31bOllama5.2 tok/s78.510:47glm-4.7-flash-unsloth:small-hotfixed-90kHEATOllama29.4 tok/s65.210:40gpt-oss:20bOllama67.8 tok/s75.9Apr 16, 2026
21:11glm-4.7-flash-unsloth:small-hotfixed-90kHEATOllama28.6 tok/s67.521:07glm-4.7-flash-unsloth:small-hotfixed-90kHEATOllama6.9 tok/s25.921:04gpt-oss:20bOllama63.5 tok/s77.921:01gpt-oss:20bOllama48.6 tok/s64.519:40glm-4.7-flash-unsloth:small-hotfixed-90kHEATOllama24.6 tok/s68.919:31ministral-3:14bOllama41.6 tok/s68.418:30gemma4:31bOllama5.5 tok/s55.018:21gemma4:26bOllama23.7 tok/s60.618:16gemma4:26bOllama15.9 tok/s21.718:11glm-4.7-flash-unsloth:small-hotfixed-90kHEATOllama25.5 tok/s53.118:06gpt-oss:20bOllama62.1 tok/s73.4