rank,name,id,score,params,min_vram_gb,license,context,benchmark_MMLU
1,Phi-4 14B,phi4-14b,69.1,14,9,MIT,16384,84.8
2,Qwen 2.5 7B,qwen25-7b,67.0,7,5,Apache 2.0,131072,74.2
3,Granite 4.1 8B Instruct,granite41-8b,66.5,8,5,Apache 2.0,131072,73.84
4,Qwen 2.5 3B Instruct,qwen25-3b,65.8,3,2,Qwen Research License,32768,65.6
5,Qwen 2.5 14B Instruct,qwen25-14b,65.7,14,9,Apache 2.0,131072,79.7
6,DeepSeek R1 671B,deepseek-r1-671b,63.0,671,400,MIT,128000,90.8
7,SmolLM3 3B,smollm3-3b,61.6,3,2,Apache 2.0,128000,59.7
8,Mistral Small 3.1 24B,mistral-small-31-24b,61.4,24,14,Apache 2.0,128000,80.6
9,Llama 3.1 8B,llama3-8b,60.5,8,6,Llama 3 Community,131072,73.0
10,Qwen 2.5 32B,qwen25-32b,60.5,32,19,Apache 2.0,131072,83.3
11,Mistral Small 3,mistral-small-24b,59.9,24,14,Apache 2.0,32768,81.0
12,Llama 3.2 3B,llama3-3b,59.5,3,2,Llama 3 Community,131072,63.4
13,Granite 3.2 8B Instruct,granite32-8b,57.5,8,5,Apache 2.0,128000,67.0
14,Llama 3.1 405B Instruct,llama-3-1-405b,57.5,405,240,Llama 3.1 Community,128000,88.6
15,Llama 3.1 70B,llama3-70b,55.9,70,40,Llama 3 Community,131072,86.0
16,Llama 3.3 70B Instruct,llama33-70b,55.9,70,40,Llama 3.3 Community,128000,86.0
17,Qwen 2.5 72B Instruct,qwen25-72b,55.9,72,42,Qwen License,131072,86.1
18,Mistral Nemo 12B Instruct,mistral-nemo-12b,55.8,12,7,Apache 2.0,128000,68.0
19,Gemma 3 27B,gemma3-27b,55.4,27,16,Gemma,128000,78.6
20,Falcon 3 10B Instruct,falcon3-10b,54.6,10,6,TII Falcon-LLM License 2.0,32000,73.1
21,Gemma 2 9B,gemma2-9b,54.3,9,6,Gemma,8192,71.3
22,Falcon 3 7B Instruct,falcon3-7b,53.1,7,5,TII Falcon-LLM License 2.0,32000,70.5
23,Phi-3.5 Mini,phi35-mini,51.8,3.8,10,MIT,131072,69.0
24,Phi-4 Mini 3.8B,phi4-mini,49.7,3.8,10,MIT,128000,67.3
25,Yi 1.5 34B Chat,yi-15-34b,48.0,34,20,Apache 2.0,4096,77.2
26,Mixtral 8x22B Instruct,mixtral-8x22b,47.4,141,82,Apache 2.0,64000,77.8
27,Gemma 2 27B,gemma2-27b,47.3,27,16,Gemma,8192,75.2
28,Mistral 7B Instruct,mistral-7b-instruct,46.4,7,5,Apache 2.0,32768,60.1
29,Gemma 2 2B,gemma2-2b,45.3,2,1,Gemma,8192,52.2
30,Falcon Mamba 7B,falcon-mamba-7b,39.9,7,5,TII Falcon-LLM License 2.0,8192,62.0
31,Mixtral 8x7B,mixtral-8x7b,39.5,47,26,Apache 2.0,32768,70.6
32,OLMoE 1B-7B Instruct,olmoe-1b-7b,34.8,7,4,Apache 2.0,4096,52.0
33,DBRX Instruct,dbrx-instruct,34.7,132,76,Databricks Open Model License,32768,73.7
34,Snowflake Arctic Instruct,arctic-instruct,30.7,480,290,Apache 2.0,4096,67.3
,Lucie 7B,lucie-7b,,7,5,Apache 2.0,4096,
,CroissantLLM 1.3B,croissant-llm,,1.3,1,MIT,2048,
,Qwen 2.5 Coder 7B,qwen25-coder-7b,,7,5,Apache 2.0,131072,
,Qwen 2.5 Coder 32B,qwen25-coder-32b,,32,19,Apache 2.0,131072,
,DeepSeek R1 Distill 7B,deepseek-r1-7b,,7,5,MIT,32768,
,DeepSeek R1 Distill 32B,deepseek-r1-32b,,32,19,MIT,32768,
,DeepSeek Coder V2 Lite 16B,deepseek-coder-v2-16b,,16,10,MIT,131072,
,Llama 3.2 Vision 11B,llama32-vision-11b,,11,8,Llama 3 Community,131072,
,Qwen 2 VL 7B,qwen2-vl-7b,,7,6,Apache 2.0,32768,
,Qwen 3 8B,qwen3-8b,,8,5,Apache 2.0,131072,
,Qwen 3 14B,qwen3-14b,,14,9,Apache 2.0,131072,
,Qwen 3 32B,qwen3-32b,,32,19,Apache 2.0,131072,
,Qwen 3 235B-A22B,qwen3-235b-a22b,,235,142,Apache 2.0,131072,
,Qwen 2.5 VL 7B,qwen25-vl-7b,,7,6,Apache 2.0,128000,
,Qwen 2.5 VL 72B,qwen25-vl-72b,,72,42,Qwen License,128000,
,Qwen 2.5 Omni 7B,qwen25-omni-7b,,7,6,Apache 2.0,32768,
,QwQ 32B,qwq-32b,,32,19,Apache 2.0,131072,
,DeepSeek R1 Distill Llama 70B,deepseek-r1-distill-llama-70b,,70,40,Llama 3.3 Community + DeepSeek,128000,
,DeepSeek V3 671B,deepseek-v3-671b,,671,400,DeepSeek License,128000,
,Gemma 3 4B,gemma3-4b,,4,10,Gemma,128000,
,Gemma 3 12B,gemma3-12b,,12,7,Gemma,128000,
,Phi-4 Multimodal 5.6B,phi4-multimodal,,5.6,4,MIT,128000,
,Phi-4 Reasoning 14B,phi4-reasoning-14b,,14,9,MIT,32768,
,Command R+ 104B (08-2024),command-r-plus-104b,,104,60,CC-BY-NC 4.0,128000,
,Aya Expanse 8B,aya-expanse-8b,,8,5,CC-BY-NC 4.0,8192,
,Aya Expanse 32B,aya-expanse-32b,,32,19,CC-BY-NC 4.0,8192,
,EuroLLM 9B Instruct,eurollm-9b,,9,6,Apache 2.0,4096,
,Teuken 7B Instruct,teuken-7b,,7,5,Apache 2.0 (commercial),4096,
,Pleias 3B Preview,pleias-3b,,3,2,Apache 2.0,2048,
,Pleias-RAG 1B,pleias-rag-1b,,1.2,0,Apache 2.0,2048,
,Moshi 7B,moshi-7b,,7.6,5,CC-BY 4.0,4096,
,Helium 1 2B,helium-1-2b,,2,1,CC-BY-SA 4.0,4096,
,SmolLM2 1.7B Instruct,smollm2-17b,,1.7,1,Apache 2.0,8192,
,SmolVLM2 2.2B Instruct,smolvlm2-22b,,2.2,1,Apache 2.0,8192,
,GLM-5.1,glm-51,,744,445,MIT,200000,
,MiniMax-M2.7,minimax-m27,,229,138,Apache 2.0,205000,
,Gemma 4 31B,gemma4-31b,,31,18,Gemma,256000,
,Gemma 4 E4B,gemma4-e4b,,4,10,Gemma,128000,
,Qwen 3.5 9B,qwen35-9b,,9,6,Apache 2.0,262000,
,Qwen 3.5 27B,qwen35-27b,,27,16,Apache 2.0,262000,
,Qwen 3.5 397B-A17B,qwen35-397b-a17b,,397,240,Apache 2.0,262000,
,Qwen 3.6 35B-A3B,qwen36-35b-a3b,,35,21,Apache 2.0,262000,
,Qwen3-Coder-Next 80B-A3B,qwen3-coder-next,,80,48,Apache 2.0,262000,
,Mistral Small 4,mistral-small-4,,119,72,Apache 2.0,256000,
,Devstral Small 2 24B,devstral-small-2,,24,14,Apache 2.0,256000,
,Voxtral-4B-TTS,voxtral-4b-tts,,4,10,CC-BY-NC 4.0,4096,
,DeepSeek R2 32B,deepseek-r2-32b,,32,19,MIT,128000,
,DeepSeek V3.2,deepseek-v32,,685,410,MIT,128000,
,Kimi K2.5,kimi-k25,,1000,600,Modified MIT,256000,
,Nemotron 3 Super 120B,nemotron-3-super-120b,,120,72,NVIDIA Open Model License,128000,
,OLMo 3 7B,olmo3-7b,,7,5,Apache 2.0,8192,
,OLMo 3 32B,olmo3-32b,,32,19,Apache 2.0,65536,
,Tiny Aya 3.35B,tiny-aya-3b,,3.35,2,CC-BY-NC 4.0,8192,
,Granite 4.0 3B Vision,granite4-3b-vision,,3,2,Apache 2.0,16384,
,Step 3.5 Flash,step-35-flash,,196,118,Apache 2.0,256000,
,Falcon H1R 7B,falcon-h1r-7b,,7,5,TII Falcon-LLM License 2.0,32768,
,Mistral Small 3.2 24B,mistral-small-32-24b,,24,14,Apache 2.0,128000,
,Codestral 22B v0.1,codestral-22b,,22,13,Mistral Non-Production License,32000,
,Codestral Mamba 7B,codestral-mamba-7b,,7,5,Apache 2.0,256000,
,Magistral Small 24B,magistral-small-24b,,24,14,Apache 2.0,128000,
,Mistral Large 3 675B,mistral-large-3,,675,405,Apache 2.0,256000,
,Mistral Medium 3.5 128B,mistral-medium-35,,128,74,Modified MIT,256000,
,Llama 4 Scout 109B,llama-4-scout,,109,65,Llama 4 Community,10000000,
,Llama 4 Maverick 400B,llama-4-maverick,,400,240,Llama 4 Community,1000000,
,Llama 3.1 Nemotron 70B,llama31-nemotron-70b,,70,40,Llama 3.1 Community,128000,
,Qwen 2.5 Coder 1.5B Instruct,qwen25-coder-15b,,1.5,1,Apache 2.0,32768,
,Qwen 2.5 Coder 3B Instruct,qwen25-coder-3b,,3,2,Qwen Research License,32768,
,Qwen 2.5 Coder 14B Instruct,qwen25-coder-14b,,14,9,Apache 2.0,131072,
,Qwen 3 30B-A3B,qwen3-30b-a3b,,30,19,Apache 2.0,131072,
,DeepSeek R1 Distill Qwen 1.5B,deepseek-r1-distill-qwen-15b,,1.5,1,MIT,131072,
,DeepSeek R1 Distill Qwen 14B,deepseek-r1-distill-qwen-14b,,14,9,MIT,131072,
,Phi-4 Mini Reasoning 3.8B,phi4-mini-reasoning,,3.8,10,MIT,128000,
,Gemma 3n E2B,gemma3n-e2b,,2,2,Gemma,32768,
,Gemma 3n E4B,gemma3n-e4b,,4,4,Gemma,32768,
,Granite 3.3 8B Instruct,granite33-8b,,8,5,Apache 2.0,128000,
,Granite 4.0 H-Small 32B-A9B,granite4-small,,32,19,Apache 2.0,128000,
,Granite 4.0 H-Tiny 7B-A1B,granite4-tiny,,7,4,Apache 2.0,128000,
,Tülu 3 8B,tulu3-8b,,8,6,Llama 3.1 Community,128000,
,Tülu 3 70B,tulu3-70b,,70,40,Llama 3.1 Community,128000,
,Molmo 7B-D,molmo-7b,,7,5,Apache 2.0,4096,
,Molmo 72B,molmo-72b,,72,42,Apache 2.0,4096,
,MiniCPM-V 2.6 8B,minicpm-v-26,,8,5,MiniCPM Model License,32000,
,MiniCPM-o 2.6 8B,minicpm-o-26,,8,5,MiniCPM Model License,32000,
,Command R 35B v01,command-r-35b,,35,20,CC-BY-NC 4.0,128000,
,Aya 23 8B,aya-23-8b,,8,5,CC-BY-NC 4.0,8192,
,Aya 23 35B,aya-23-35b,,35,20,CC-BY-NC 4.0,8192,
,Yi Coder 9B Chat,yi-coder-9b,,9,5,Apache 2.0,128000,
,Jais 30B Chat v3,jais-30b,,30,18,Apache 2.0,8192,
,Jais Adapted 70B Chat,jais-70b,,70,40,Apache 2.0,4096,
,Sarvam-M 24B,sarvam-m-24b,,24,14,Apache 2.0,32768,
,Salamandra 7B Instruct,salamandra-7b,,7.7,5,Apache 2.0,8192,
,Salamandra 40B Instruct,salamandra-40b,,40,24,Apache 2.0,8192,
,EuroLLM 22B Instruct 2512,eurollm-22b,,22.6,13,Apache 2.0,32768,
,Claire 7B 0.1,claire-7b,,7,5,CC-BY-NC-SA 4.0,2048,
,Jamba 1.5 Mini,jamba-15-mini,,52,30,Jamba Open Model License,256000,
,Hunyuan-A13B Instruct,hunyuan-a13b,,80,48,Tencent Hunyuan License,262144,
,LLaVA-OneVision 7B,llava-onevision-7b,,7,5,Apache 2.0,32768,
,LLaVA-OneVision 72B,llava-onevision-72b,,72,42,Apache 2.0,32768,
,Grok-1 (base),grok-1,,314,188,Apache 2.0,8192,
,gpt-oss 120B,gpt-oss-120b,,117,70,Apache 2.0,128000,
,gpt-oss 20B,gpt-oss-20b,,21,13,Apache 2.0,128000,
,Kimi K2.6,kimi-k26,,1000,600,Modified MIT,256000,
,Qwen 3 VL 235B-A22B,qwen3-vl-235b,,235,142,Apache 2.0,262144,
,Qwen 3 VL 30B-A3B,qwen3-vl-30b,,30,19,Apache 2.0,262144,
,Qwen 3 VL 8B,qwen3-vl-8b,,8,6,Apache 2.0,262144,
,ERNIE 4.5 300B-A47B,ernie-45-300b,,300,180,Apache 2.0,131072,
,ERNIE 4.5 21B-A3B Thinking,ernie-45-21b,,21,13,Apache 2.0,131072,
,Ring-1T,ring-1t,,1000,600,MIT,131072,
,Seed-OSS 36B Instruct,seed-oss-36b,,36,22,Apache 2.0,524288,
,EXAONE 4.5 33B,exaone-45-33b,,33,20,EXAONE AI Model License,262144,
,Nemotron Nano 3 30B-A3B,nemotron-nano-3-30b,,30,19,NVIDIA Open Model License,1000000,
,Nemotron Nano v2 VL 12B,nemotron-nano-v2-vl-12b,,12.6,8,NVIDIA Open Model License,128000,
,Apertus 70B,apertus-70b,,70,40,Apache 2.0,65536,
,Apertus 8B,apertus-8b,,8,6,Apache 2.0,65536,
,Trinity Mini 26B-A3B,trinity-mini-26b,,26,15,Apache 2.0,131072,
,Hunyuan Large 2.0,hunyuan-20-large,,406,245,Tencent Hunyuan License,262144,
,InternVL 3.5 8B,internvl-35-8b,,8,6,Apache 2.0,32768,
,MiMo V2 Flash,mimo-v2-flash,,309,185,MIT,128000,
,Rakuten AI 3.0,rakuten-ai-3,,700,420,Apache 2.0,32768,
,Kanana 2 30B-A3B Thinking,kanana-2-30b,,30,18,Apache 2.0,131072,
,DeepSeek-OCR,deepseek-ocr,,3,2,MIT,8192,
,HunyuanOCR 1B,hunyuan-ocr-1b,,1,0,Tencent Hunyuan License,8192,
,Gemma 4 26B-A4B MoE,gemma4-26b-moe,,26,16,Gemma,128000,
,dots.llm1 Instruct,dots-llm1,,142,85,MIT,32768,
,Qwen 3 Omni 30B-A3B,qwen3-omni-30b,,30,19,Apache 2.0,131072,
,Qwen 3.5 122B-A10B,qwen35-122b-a10b,,122,73,Apache 2.0,262000,
,Pangu Pro MoE 72B,pangu-pro-moe-72b,,72,42,Pangu Model License,32768,
,Qwen 3.6 27B,qwen36-27b,,27,16,Apache 2.0,262144,
,DeepSeek V4 Pro 1.6T,deepseek-v4-pro,,1600,960,MIT,1000000,
,DeepSeek V4 Flash 284B,deepseek-v4-flash,,284,170,MIT,1000000,
,Tencent Hy3 Preview 295B,tencent-hy3-preview,,295,177,Tencent Hunyuan License,256000,
,LLaDA 2.0 Uni 16B,llada2-uni,,16,18,Apache 2.0,8192,
,MiMo V2.5 Pro,mimo-v25-pro,,1020,595,MIT,1000000,
,MiMo V2.5,mimo-v25,,310,180,MIT,1000000,
,Nemotron 3 Nano Omni 30B-A3B,nemotron-omni-30b,,30,21,NVIDIA Open Model License,256000,
,Laguna XS.2,laguna-xs2,,33,19,Apache 2.0,131072,
,Granite 4.1 30B Instruct,granite41-30b,,30,17,Apache 2.0,131072,
,Granite 4.1 3B Instruct,granite41-3b,,3,2,Apache 2.0,131072,
,Ling 2.6 1T,ling-26-1t,,1000,580,MIT,262144,
,Gemma 4 E2B,gemma4-e2b,,2,7,Gemma,128000,
,Nemotron Cascade 2 30B-A3B,nemotron-cascade-2,,30,17,NVIDIA Open Model License,128000,
,Nemotron 3 33B,nemotron3,,33,19,NVIDIA Open Model License,128000,
,Nemotron 3 Nano 30B-A3B,nemotron-3-nano,,30,17,NVIDIA Open Model License,128000,
,MedGemma 4B,medgemma,,4,2,Gemma,128000,
,Gemma 4 2B,gemma4,,2,1,Gemma,128000,
,Qwen 3.5 0.8B,qwen3-5,,0.8,0,Apache 2.0,256000,
,MedGemma 1.5 4B,medgemma1-5,,4,2,Gemma,128000,
,Granite 4.1,granite4-1,,3,1,Apache 2.0,128000,
,Qwen 3.6 27B,qwen3-6,,27,16,Qwen License,256000,
,LFM2.5 Thinking 1.2B,lfm2-5-thinking,,1.2,0,LFM Open License v1.0,32768,
,GLM 4.7 Flash,glm-4-7-flash,,3,1,MIT,128000,