Machine
RTX A6000 · Core i9-14900K #294df0f1
48 runs of 27 models since Oct 20, 2026. Last run Nov 5, 2026.
48
runs
27
models
33
configs
11
configs run 2+ times
Hardware
- GPU
- NVIDIA RTX A6000 47 GB
- CPU
- Intel(R) Core(TM) i9-14900K 24 cores
- RAM
- 63 GB
- OS
- Windows 11 רҵ�� 25H2
- Tools
- llama.cpp 48
- Client
- 0.4.98
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, 27 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
Qwen3.8-27B-GSQ-RCO-IQ3_S-mtp
1 run · 1 config
LuffyTheFox-Swift-Qwen3.8-27B-Genesis-GGUF-NVFP4
1 run · 1 config
JonathanColetti-Qwen3.8-27B-Uncensored-GGUF-Q5_K_M
3 runs · 1 config
★ bestQ5_K_Mllama.cpp256k ctxthinking oneffort xhigh
×3 runs
87.785.1–90.5
steady1 fast run
30.6 tok/s
30.0–50.8
Qwen3.8-27B-EfficientThink-Uncensored-K3-Opus5-Grok4.6-GPT5.6Sol-SFT-SimPO-DFlash2
3 runs · 1 config
★ bestllama.cpp256k ctxthinking oneffort xhigh
×3 runs
87.183.8–88.7
steady
26.4 tok/s
26.3–26.9
Qwen3.8-27B-Omnimerge-v6-Q4_K_L
1 run · 1 config
Qwen3.8-27B-Omnimerge-v6-MTP-GGUF
2 runs · 2 configs
Qwen3.8-27B-EfficientThink-Uncensored-K3-Opus5-Grok4.6-GPT5.6Sol-SFT-SimPO-DFlash2-GGUF-Q6_K
1 run · 1 config
Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF-RVN-Q5_K_M-mtp
4 runs · 2 configs
★ bestQ5_K_Mllama.cpp256k ctxthinking oneffort xhigh
×2 runs
85.985.8–86.0
steady
30.0 tok/s
29.7–30.2
Qwen3.8-27B-pi-GGUF-Q4_K_M
1 run · 1 config
Qwen3.8-27B-Cold-Fusion-GAIN-V1.1-NM-DAU-NEO-MAX-MTP-GGUF-Q6_K
2 runs · 1 config
★ bestQ6_Kllama.cpp256k ctxthinking oneffort xhigh
×2 runs
85.783.0–88.4
steady
34.2 tok/s
24.7–43.7
Compare with another machine
Run history
Nov 4, 2026
03:20Swift-Qwen3.8-27B-GGUF-Q6_Kllama.cpp24.7 tok/s80.302:45Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF-RVN-Q6_K-multilingual-mtpllama.cpp27.3 tok/s89.501:59Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF-RVN-Q5_K_M-mtpllama.cpp30.2 tok/s86.0Nov 3, 2026
09:17Qwen3.8-27B-GSQ-RCO-IQ3_S-mtpllama.cpp39.4 tok/s88.508:14Qwen3.8-27B-EfficientThink-Uncensored-K3-Opus5-Grok4.6-GPT5.6Sol-SFT-SimPO-DFlash2llama.cpp26.9 tok/s88.705:36Qwen3.8-27B-Cold-Fusion-GAIN-V1.1-NM-DAU-NEO-MAX-MTP-GGUF-Q6_Kllama.cpp24.7 tok/s83.001:23Swift-Qwen3.8-27B-GGUF-Q6_Kllama.cpp25.1 tok/s84.500:40Swift-Qwen3.8-27B-Genesis-NVFP4-v4llama.cpp35.0 tok/s82.0Nov 2, 2026
09:13JonathanColetti-Qwen3.8-27B-Uncensored-GGUF-Q5_K_Mllama.cpp30.6 tok/s85.108:34Qwen3.8-27B-Omnimerge-v6-Q4_K_Lllama.cpp33.4 tok/s87.007:41Qwen3.8-27B-Coder390--MTP-Q4_K_Mllama.cpp30.4 tok/s79.107:11Qwen3.8-27B-pi-GGUF-Q4_K_Mllama.cpp33.9 tok/s85.9Oct 30, 2026
07:58Qwen3.8-27B-Omnimerge-v6-MTP-GGUFllama.cpp44.7 tok/s83.007:08Qwen3.8-27B-Coder390--MTPllama.cpp47.2 tok/s82.206:11Qwen3.8-27B-Omnimerge-v6-MTP-GGUFllama.cpp56.6 tok/s86.802:07Qwen3.8-27B-pi-GGUFllama.cpp33.7 tok/s84.3Show all 48 runs ▾Show fewer ▴
Oct 28, 2026
09:15Qwen3.8-27B-UD-Q8_K_XL×2llama.cpp20.6 tok/s81.005:40Qwen3.8-27B-UD-Q6_K_XL×2llama.cpp23.8 tok/s84.502:42Swift-Qwen3.8-27B-GGUF-Q6_Kllama.cpp24.5 tok/s82.101:50Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF-RVN-Q6_K-multilingual-mtp×2llama.cpp27.3 tok/s83.4Oct 27, 2026
07:49Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF-RVN-Q6_K-multilingual-mtpllama.cpp27.4 tok/s86.801:16Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF-RVN-Q5_K_M-mtp×2llama.cpp29.7 tok/s83.4Oct 26, 2026
07:52Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF-RVN-Q5_K_M-mtpllama.cpp29.7 tok/s85.804:17JonathanColetti-Qwen3.8-27B-Uncensored-GGUF-Q5_K_Mllama.cpp30.0 tok/s90.503:27LuffyTheFox-Swift-Qwen3.8-27B-Genesis-GGUF-NVFP4llama.cpp36.6 tok/s88.5Oct 23, 2026
09:39Qwen3.8-27B-EfficientThink-Uncensored-K3-Opus5-Grok4.6-GPT5.6Sol-SFT-SimPO-DFlash2-GGUF-Q6_Kllama.cpp42.6 tok/s86.809:10Qwen3.8-27B-Cold-Fusion-GAIN-V1.1-NM-DAU-NEO-MAX-MTP-GGUF-Q6_Kllama.cpp43.7 tok/s88.407:56Swift-1.5-Qwen3.8-Flash-Next-GSQ-RCO-abliterated-GGUFllama.cpp38.2 tok/s74.505:22SC117-Swift-IQ3_S×2llama.cpp88.1 tok/s85.603:19SC117-Swift-IQ3_Sllama.cpp109 tok/s73.401:14Scion-35B-A3B-PQ2_0-corrllama.cpp101 tok/s73.1Oct 22, 2026
08:30G9v3-3B-GGUFllama.cpp142 tok/s48.408:15NeoHorse-1-4B-GGUFllama.cpp106 tok/s72.707:59NeoHorse-1-4B-GGUFllama.cpp106 tok/s76.5