Machine
2× RTX 5060 · Core i7-8700 #3a7b23fa
199 runs of 100 models since Sep 8, 2026. Last run Nov 21, 2026.
199
runs
100
models
118
configs
41
configs run 2+ times
Hardware
- GPU
- 2× NVIDIA GeForce RTX 5060 16 GB
- CPU
- Intel(R) Core(TM) i7-8700 CPU @ 3.20GHz 6 cores
- RAM
- 31 GB
- OS
- Ubuntu 26.04.1 LTS
- Tools
- llama.cpp 191 · capy-flash 5 · PolyStrata 3
- Client
- 0.4.99
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, 100 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-EfficientThink-VAL-APEX-I-NanoPlus-V3
1 run · 1 config
Qwen3.8-27B-GSQ-RCO-IQ3_S
1 run · 1 config
Qwen3.8-27B-i1-IQ4_XS-GGUF-Smaller
1 run · 1 config
Dirk-Qwen3.8-27B-UD-IQ4_XS
4 runs · 1 config
★ bestllama.cpp64k ctxthinking on
×4 runs
85.982.6–87.8
steady
37.5 tok/s
36.6–38.2
ornith-1.5-35b-a3b-apex-i-quality
1 run · 1 config
VeriLoop-E2-GSQ-RCO-IQ3_S-mtp
1 run · 1 config
Qwen3.8-27B-GSQ-RCO-IQ3_S-recovered
1 run · 1 config
Qwen3.8-27B-4.2BPW-16GB
1 run · 1 config
Qwen3.8-27B-YMQ-M-TI
6 runs · 2 configs
★ bestllama.cpp96k ctxthinking on
×2 runs
85.483.5–87.2
steady
36.8 tok/s
36.8–36.9
Tiel-Coder-35B-A3B.APEX-I-MiniPlus-V2.1
1 run · 1 config
Compare with another machine
Run history
Nov 15, 2026
18:28qwen3.8-27b-gsq3capy-flash116 tok/s82.0Nov 9, 2026
11:39qwen3.6-35b-a3b-ud-q4_k_mPolyStrata129 tok/s80.611:03tiel-coder-35b-a3b-ud-q4_k_xlPolyStrata109 tok/s84.810:28ornith-1.5-35b-a3b-apex-i-qualityPolyStrata93.2 tok/s85.9Show all 199 runs ▾Show fewer ▴
Oct 27, 2026
20:50qwen3.8-27b-gsq3capy-flash114 tok/s83.920:17qwen3.8-27b-gsq3capy-flash115 tok/s85.217:20Ornith-1.5-35B-Q4_K_Mllama.cpp46.8 tok/s81.711:57qwen3.8-27b-gsq3capy-flash115 tok/s83.3Oct 24, 2026
21:59qwen3.8-27b-gsq3capy-flash114 tok/s83.9Oct 22, 2026
14:46Qwen3.8-27B-BaseQuant-V1-Smallllama.cpp40.1 tok/s79.608:08raon-1-mini-preview-Q4_K_Mllama.cpp43.7 tok/s80.900:45flint-UD-Q4_K_XLllama.cpp45.8 tok/s80.8Oct 20, 2026
20:45Ornith-1.5-35B-Q4_K_Mllama.cpp53.2 tok/s86.819:40Ornith-1.5-35B-Q4_K_Mllama.cpp56.5 tok/s88.7Oct 19, 2026
08:17MiniCPM-V-4.7-35B-A3B.APEX-I-NanoPlus-V2.1×2llama.cpp97.7 tok/s76.700:29Cyber-Tiel-Coder-35B-A3B-UD-IQ3_XXSllama.cpp77.5 tok/s82.3Oct 11, 2026
21:42HERO-35B-A3B.i1-IQ3_XXSllama.cpp83.7 tok/s65.917:38Ternary-Bonsai-2-27B-Q2_0-prism-fork-requiredllama.cpp38.4 tok/s83.400:08Qwen3.8-27B-VAL-APEX-I-NanoPlus-V3llama.cpp36.5 tok/s84.1Oct 8, 2026
21:09VeriLoop-E2-GSQ-RCO-IQ3_S-mtpllama.cpp39.7 tok/s85.811:53Underdog-Saluki-27B-1.0-IQ2-mix-MTPllama.cpp39.4 tok/s82.009:35AREX-2.i1-Q3_K_Sllama.cpp37.2 tok/s43.308:14Signal-3.8-27B-Terse-Coder.i1-Q3_K_Sllama.cpp37.3 tok/s75.3Oct 5, 2026
19:22gpt-oss-20b-UD-Q8_K_XL×2llama.cpp101 tok/s70.011:51Qwen3.8-27B-YMQ-M-TI×2llama.cpp36.8 tok/s85.401:21Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXS×4llama.cpp77.3 tok/s84.5Oct 4, 2026
22:50Qwen3.8-27B-GSQ-RCO-IQ3_S-recoveredllama.cpp25.6 tok/s85.722:01Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXS×2llama.cpp78.5 tok/s84.621:24Mellum2-12B-A2.5B-Thinking-Q8_0llama.cpp115 tok/s71.320:45Ornith-1.5-35B-A3B-GSQ-RCO-3bitllama.cpp85.5 tok/s80.401:06Whittle-Next-27B-A3B-v4.4-Q3_K_Mllama.cpp75.1 tok/s65.000:29Qwen3.6-35B-A3B-GSQ-hybridllama.cpp81.0 tok/s82.6Oct 3, 2026
23:39Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6Kllama.cpp34.6 tok/s81.822:01Qwen3.6-35B-A3B-GSQ-hybrid×2llama.cpp83.8 tok/s77.920:15Qwen3.8-27B-YMQ-XS-TIllama.cpp40.2 tok/s79.9Oct 2, 2026
21:37Ornith-1.5-35B-A3B-GSQ-RCO-3bitllama.cpp85.4 tok/s85.121:18FrogNano-4B-2609-Q6_Kllama.cpp90.4 tok/s72.518:40Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXS×5llama.cpp76.7 tok/s86.411:03Qwen3.8-27B-GSQ-RCO-IQ3_Sllama.cpp25.6 tok/s87.609:48Qwen3.8-27B-GSQ-RCO-IQ3_XXSllama.cpp39.2 tok/s82.8Oct 1, 2026
23:06apodex_Apodex-1.1-mini-IQ2_Mllama.cpp85.0 tok/s53.322:58apodex_Apodex-1.1-mini-IQ2_Mllama.cpp103 tok/s80.422:27Apodex-1.1-mini.i1-IQ3_Mllama.cpp62.4 tok/s79.721:41apodex_Apodex-1.1-mini-IQ2_Mllama.cpp103 tok/s67.321:20apodex_Apodex-1.1-mini-IQ2_Mllama.cpp86.3 tok/s62.618:41Qwen3.8-27B-YMQ-M-TIllama.cpp35.3 tok/s81.718:29Qwen3.8-27B-Ridge-3.7bpwllama.cpp34.6 tok/s81.217:59Qwen3.8-27B-YMQ-M-TI×3llama.cpp35.7 tok/s84.816:43Nex-N2.5-mini-IQ3_XXS×2llama.cpp80.7 tok/s84.312:34Signal-3.8-27B-AP-IQ3_XXS×2llama.cpp26.5 tok/s83.0Sep 30, 2026
23:59gpt-oss-20b-UD-Q6_K_XLllama.cpp101 tok/s71.923:33gpt-oss-20b-MXFP4llama.cpp95.6 tok/s74.022:30gpt-oss-20b-nvfp4llama.cpp97.7 tok/s66.421:51Nex-N2.5-mini-mixed-Q2Q3-vision-trial×3llama.cpp88.8 tok/s82.006:51KAT-Coder-V2.5-Dev.i1-IQ3_XXSllama.cpp84.6 tok/s76.506:44Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXSllama.cpp77.5 tok/s80.601:31occamy-1.0.i1-IQ3_XXSllama.cpp80.2 tok/s85.200:38Nex-N2.5-mini-IQ3_XXSllama.cpp79.6 tok/s82.900:14Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXSllama.cpp76.7 tok/s81.2Sep 29, 2026
23:59Tiel-Coder-35B-A3B.APEX-I-NanoPlusllama.cpp81.4 tok/s67.223:07Occamy-1.0.APEX-I-MiniPlus-V2.1llama.cpp75.8 tok/s62.523:00occamy-1.0.i1-IQ3_XXS×2llama.cpp77.2 tok/s57.422:02Occamy-1.0.APEX-I-MiniPlus-V2.1llama.cpp76.9 tok/s—08:31Sharp-MiniCPM5-2B-Q8_K_Mllama.cpp127 tok/s55.8Sep 28, 2026
19:11Whittle-Qwen-3.8-35B-A3B-Q3_K_Mllama.cpp76.6 tok/s44.811:51Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXSllama.cpp89.7 tok/s86.411:09Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL×2llama.cpp93.6 tok/s74.509:29Swift-1.5-Qwen3.8-27B-GSQ-RCO-IQ3_XXS-mtp×2llama.cpp39.2 tok/s76.308:06Muse-Glimmer-30B-AK-Q3_K_XLllama.cpp25.0 tok/s83.302:03KAT-Coder-V2.5-Dev.i1-IQ3_XXSllama.cpp86.9 tok/s77.101:48Unsloth-Ornith-1.5-35B-A3B-UD-IQ3_XXSllama.cpp89.7 tok/s81.001:12Nex-N2.5-mini-APEX-Mini×2llama.cpp88.2 tok/s81.1Sep 27, 2026
21:59Nex-N2.5-mini.APEX-I-MiniPlus-V2.1llama.cpp74.0 tok/s83.821:00Qwen3.8-27B-OrcaRouter-GSQ-RCO-IQ3_XXS-v2.0-qatfallama.cpp36.5 tok/s81.820:02Tiel-Coder-35B-A3B.APEX-I-NanoPlusllama.cpp84.9 tok/s79.9Sep 25, 2026
20:14Ornith-1.5-35B-A3B-GSQ-RCO-3bit×2llama.cpp103 tok/s84.316:41Unsloth-Ornith-1.5-35B-A3B-UD-IQ3_XXSllama.cpp88.8 tok/s83.212:41Qwen3.8-27B-UD-IQ4_XSllama.cpp25.1 tok/s84.811:59Swift-1.5-Qwen3.8-27B-GSQ-RCO-IQ3_XXS-mtpllama.cpp39.3 tok/s86.810:30Swift-1.5-Qwen3.8-27B-GSQ-RCO-IQ3_S-mtp×2llama.cpp37.9 tok/s82.4Sep 24, 2026
20:28OliviaRossi_MiMo-Ornith-9B-AGSI-Q6_Kllama.cpp83.9 tok/s79.001:43Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXSllama.cpp112 tok/s82.200:18Qwen3.8-35B-A3B-Distill-APEX-Minillama.cpp83.4 tok/s74.6Sep 23, 2026
23:34Cyber-Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXSllama.cpp117 tok/s70.422:32Ornith-1.5-9B-Q6_Kllama.cpp74.4 tok/s81.1Sep 22, 2026
16:17Ornith-1.5-9B-Q6_Kllama.cpp52.9 tok/s68.713:55MiMo-V2.6-Distill-Qwen-9B-Q6_K×2llama.cpp58.7 tok/s70.300:31Ornith-1.5-9B-Q6_Kllama.cpp52.8 tok/s80.9Sep 21, 2026
14:12Qwen3.8-27B-Q3_Kllama.cpp25.7 tok/s81.608:06Qwen3.8-35B-A3B-Distill.APEX-I-MiniPlus-V2.1llama.cpp85.0 tok/s78.600:52Ternary-Bonsai-2-27B-PQ2_0llama.cpp41.8 tok/s43.8Sep 20, 2026
22:50MiniCPM5-2B-bf16llama.cpp76.1 tok/s54.2Sep 18, 2026
09:04Ternary-Bonsai-2-27B-PQ2_0llama.cpp41.4 tok/s78.500:20Ornith-1.5-9B-NVFP4-Q5K-mtpllama.cpp79.2 tok/s78.2Sep 17, 2026
21:10Qwen3.5-9B-The-Defiant-Fable-Uncnr-Heretic-plusIQ-TOOLS-NEO-MAX-MTP-Q6_Kllama.cpp43.0 tok/s75.609:17Qwen3.8-9B-Q8_0llama.cpp69.0 tok/s77.609:04Qwen3.8-4B-BF16llama.cpp62.9 tok/s72.108:32Dirk-Qwen3.8-27B-UD-IQ4_XSllama.cpp36.6 tok/s84.708:07Qwen3.8-27B-UD-Q3_K_XL×2llama.cpp36.3 tok/s83.506:16Ornith-1.5-35B-A3B-APEX-i-minillama.cpp113 tok/s77.0Sep 16, 2026
12:05Qwen3.8-27B-GSQ-RCO-IQ3_XXS-mtpllama.cpp38.8 tok/s85.207:43Maestro1-9B.Q6_Kllama.cpp55.3 tok/s67.800:03granite-4.2-3b-Q8_0llama.cpp89.6 tok/s59.1Sep 15, 2026
16:07Nex-N2.5-mini-mixed-Q2Q3-vision-trialllama.cpp103 tok/s86.215:49vectionlabs_Salience-27B-R6-IQ3_XXSllama.cpp44.7 tok/s79.815:20Maestro-2-9B-Preview.Q6_Kllama.cpp79.0 tok/s70.113:38Unsloth-Ornith-1.5-35B-A3B-UD-IQ3_XXS×2llama.cpp88.5 tok/s74.610:35Qwen3.8-27B-i1-IQ4_XS-GGUF-Smallerllama.cpp40.1 tok/s86.5Sep 14, 2026
15:11Qwen3.8-27B-IQ3_XSllama.cpp25.9 tok/s75.1Sep 13, 2026
23:02Qwen3.8-27B-IQ3_XXSllama.cpp43.5 tok/s79.722:14Unsloth-Ornith-1.5-35B-A3B-UD-IQ3_XXSllama.cpp87.7 tok/s87.122:02Unsloth-Ornith-1.5-35B-A3B-UD-IQ3_XXSllama.cpp87.2 tok/s82.321:58Unsloth-Ornith-1.5-35B-A3B-UD-IQ3_XXSllama.cpp88.5 tok/s82.4Sep 12, 2026
23:36Nail-Qwen3.6-35B-A3B-MTP-UD-IQ3_XXSllama.cpp90.6 tok/s75.115:38Muse-Glimmer-30B-UD-Q3_K_XLllama.cpp25.8 tok/s74.512:34Qwen3.8-27B-GSQ-RCO-IQ3_S-mtpllama.cpp27.1 tok/s84.503:16gemma-4-26B-A4B-it-UD-IQ4_NL×3llama.cpp78.2 tok/s74.3Sep 11, 2026
23:47Dirk-Qwen3.8-27B-UD-IQ4_XSllama.cpp38.1 tok/s82.622:55Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXSllama.cpp113 tok/s80.022:45Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XL×2llama.cpp118 tok/s78.619:52Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXSllama.cpp113 tok/s83.014:16gemma-4-26B-A4B-it-UD-IQ3_XXSllama.cpp87.6 tok/s77.412:14Qwen3.8-27B-Ridge-3.7bpwllama.cpp42.6 tok/s72.809:02Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXSllama.cpp116 tok/s86.6Sep 10, 2026
17:43Tiel-Coder-35B-A3B-UD-IQ3_XXSllama.cpp93.0 tok/s80.317:22Cyber-Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXSllama.cpp114 tok/s82.212:50Qwen3.8-27B-GSQ-RCO-IQ3_XXS-mtpllama.cpp40.0 tok/s82.2Sep 9, 2026
23:18granite-4.2-8b-Q4_K_Mllama.cpp84.2 tok/s68.522:34gpt-oss-20b-UD-Q8_K_XLllama.cpp176 tok/s77.222:16gpt-oss-20b-MXFP4llama.cpp144 tok/s77.020:49Dirk-Qwen3.8-27B-UD-IQ4_XSllama.cpp38.2 tok/s87.120:33Qwen3.8-27B-UD-IQ4_XSllama.cpp38.1 tok/s79.019:06Cyber-Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXS×2llama.cpp117 tok/s81.206:51Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXSllama.cpp115 tok/s86.3Sep 8, 2026
20:40Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XLllama.cpp98.2 tok/s84.019:29Tiel-Coder-35B-A3B-MTP-UD-Q2_K_XLllama.cpp98.0 tok/s85.713:50gpt-oss-20b-UD-Q6_K_XLllama.cpp179 tok/s70.413:32Dirk-Qwen3.8-27B-UD-IQ4_XSllama.cpp37.0 tok/s87.812:50gpt-oss-20b-UD-Q8_K_XL×2llama.cpp135 tok/s74.812:25gpt-oss-20b-UD-Q6_K_XLllama.cpp124 tok/s77.512:23gpt-oss-20b-UD-Q4_K_XL×2llama.cpp124 tok/s64.711:59gpt-oss-20b-Q4_K_M×2llama.cpp133 tok/s64.910:46gemma-4-E4B-it-qat-UD-Q4_K_XLllama.cpp100 tok/s71.510:42Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXSllama.cpp43.7 tok/s83.110:35Tiel-Coder-35B-A3B-MTP-UD-IQ3_XXSllama.cpp112 tok/s81.710:31Qwen3.8-27B-UD-IQ3_Sllama.cpp42.9 tok/s81.910:22Qwen3.8-9B-Q8_0llama.cpp73.5 tok/s75.910:17Qwen3.8-9B-Q6_Kllama.cpp83.4 tok/s72.310:07granite-4.2-8b-Q4_K_L×2llama.cpp71.6 tok/s70.409:50qwythos-9b-nvfp4llama.cpp135 tok/s50.009:40gemma-4-12B-it-qat-UD-Q4_K_XLllama.cpp50.2 tok/s70.309:31Ornith-1.5-9B-Q8_0llama.cpp63.5 tok/s76.309:18Ornith-1.5-9B-Q6_Kllama.cpp50.5 tok/s63.009:09Ornith-1.5-9B-Q8_0llama.cpp71.9 tok/s61.2