Models
61 of 61 models•
Benchmarks
Qwen 3.8 27B80.578%72%86%00
0
0
DeepSeek V4 Flash 073178.985%72%83%00
0
0
Laguna S 2.175.156%71%78%00
0
0
Qwen3.6 27B73.799%65%78%00
0
0
Qwen3.6 35B A3B71.199%65%73%00
0
0
DeepSeek V4 Flash70.487%59%77%00
0
0
Step 3.7 Flash70.0100%61%74%00
0
0
Ornith 1.0 397B67.797%64%69%00
0
0
Ornith 1.0 35B67.296%60%70%00
0
0
rNex N2 Mini60.986%53%65%00
0
0
Gemma 4 31B It59.569%67%55%00
0
0
Nemotron 3 Super 120B A12B58.869%63%56%00
0
0
Agents A157.195%56%57%00
0
0
Nemotron 3.5 Lightning56.265%58%55%00
0
0
Gpt Oss 120B56.272%55%56%00
0
0
Gemma 4 26B A4B53.680%51%54%00
0
0
Bonsai 27B53.269%50%54%00
0
0
Holo 3.1 35B A3B53.097%49%54%00
0
0
Gemma 4 12B It51.982%62%45%00
0
0
Laguna XS 2.151.858%50%53%00
0
0
Qwen3.5 9B Base51.099%50%50%00
0
0
Gpt Oss 20B48.078%50%46%00
0
0
Qwen3.5 9B46.399%49%43%00
0
0
Diffusiongemma 26B A4B It44.166%40%46%00
0
0
Mimo V2 Flash38.488%30%42%00
0
0
Nemotron 3 Nano 30B A3B38.239%41%37%00
0
0
Nemotron 3 Nano Omni 30B A3B Reasoning32.950%33%32%00
0
0
Gemma 4 E4B It28.914%28%30%00
0
0
Gemma 4 E2B It21.38%23%21%00
0
0
LFM2.5 8B A1B13.195%8%13%00
0
0
In Progress
AIBTL-4—–––00
0
0
Gemma 4 12B Coder Fable5 Composer2.5 V1—–––00
0
0
Llama 3.1 70B—–––00
0
0
Llama 3.1 70B Instruct—–––00
0
0
Llama 3.3 70B Instruct—–––00
0
0
Nvidia Nemotron 3 Ultra 550B A55B—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dflash—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dflash-n3—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dflash1—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dflash2—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dflash4—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dflash5—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dflash6—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dspark—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dspark1—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dspark4—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dspark5—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dspark6—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dspark7—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-dspark8—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-mtp2—–––00
0
0
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4-mtp4—–––00
0
0
Qwen2 72B Instruct—–––00
0
0
Qwen2.5 32B—–––00
0
0
Qwen2.5 72B—–––00
0
0
Qwen2.5 72B Instruct—–––00
0
0
Qwen2.5 Coder 32B—–––00
0
0
Qwen3 235B A22B—–––00
0
0
Qwen3-32B—–––00
0
0
Qwen3.5 122B A10B—–––00
0
0
qwen3.8-27b-ud-q4km—–––00
0
0
61 models · 195 scored variants ·Intelligence = 2% × τ²-bench + 36% × GAIA + 62% × GDPval · Decode/Prefill = measured tok/s at a cold 32k-token context, single stream · benchmark scores = best measured run, whole percents.
Intelligence / Speed →the ranked leaderboardHardware →what your hardware runs