Machine

Advanced Micro Devices Inc Ati Hawkpoint1 · Ryzen 7 H 255 w/ Radeon 780M Graphics #55a8f2cc

16 runs of 8 models since Nov 15, 2026. Last run Nov 21, 2026.

ShareRedditX
16
runs
8
models
10
configs
3
configs run 2+ times

Hardware

GPU
Advanced Micro Devices, Inc. [AMD/ATI] HawkPoint1 32 GB
CPU
AMD Ryzen 7 H 255 w/ Radeon 780M Graphics 8 cores
RAM
14 GB
OS
Ubuntu 26.04.1 LTS
Tools
llama.cpp 16
Client
0.4.99
All Advanced Micro Devices, Inc. [AMD/ATI] HawkPoint1 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 read405060708090100102050tok/s (log scale) →quality →outliers sit at the edges

Medians of each model's best config, 8 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.

Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.7BPW
1 run · 1 config
★ bestllama.cpp128k ctxthinking oneffort medium
85.2
1 rated run
9.4 tok/s
Qwen3.8-27B-GSQ-RCO-Intel-Arc-Tuned-MTP-3.2BPW
4 runs · 2 configs
★ bestllama.cpp128k ctxthinking oneffort xhigh
×3 runs
84.883.1–85.2
steady
19.0 tok/s
18.4–19.2
Ornith-1.5-9B-MTP-Q8_0
1 run · 1 config
★ bestQ8_0llama.cpp128k ctxthinking on
84.2
1 rated run
31.2 tok/s
Swift-1.5-Ornith-9B-Q8_0
1 run · 1 config
★ bestQ8_0llama.cpp128k ctxthinking on
80.0
1 rated run
32.1 tok/s
LFM2.5-24B-A2B-QAD-Q4_0
1 run · 1 config
★ bestQ4_0llama.cpp128k ctxthinking on
75.9
1 rated run
64.9 tok/s
gemma-4-26B-A4B-it-qat-UD-Q4_K_XL
6 runs · 2 configs
★ bestQ4_Kllama.cpp128k ctxthinking on
×3 runs
70.568.4–74.3
steady
30.0 tok/s
25.0–30.3
LFM2.5-24B-A2B-QAD-Q2_K_L
1 run · 1 config
★ bestQ2_Kllama.cpp128k ctxthinking on
59.4
1 rated run
65.7 tok/s
LFM3-Omni-2B-Q8_0
1 run · 1 config
★ bestQ8_0llama.cpp128k ctxthinking on
42.5
1 rated run
115 tok/s

Compare with another machine

Run history