19 Qwen3.8 Max (2.4T-A95B) Alibaba 75±1 local ~380GB 262k llama.cpp › copy 24 Kimi K3 Moonshot AI 74±1 local ~884GB 1M llama.cpp › copy 29 GLM-5.3 Z.ai 72±1 local ~239GB 1M llama.cpp › copy 33 DeepSeek V4.1 Flash DeepSeek 71±1 local ~537GB 1M how to run › copy 34 DeepSeek V4 Pro (0813) DeepSeek 71±1 local ~1010GB 1M llama.cpp › copy 40 DeepSeek V4 Flash (0731) DeepSeek 70±1 local ~92GB 1M llama.cpp › copy 42 Kimi K2.6 Moonshot AI 70±1 local ~317GB 262k llama.cpp › copy 43 GLM-5.2 Z.ai 69±1 local ~121GB 1M llama.cpp › copy 44 A.X K2 (688B-A33B) SK Telecom 69±5 * † local ~1454GB 262k how to run › copy 47 DeepSeek V4 Flash (Preview, Apr 2026) DeepSeek 69±1 local ~103GB llama.cpp › copy 49 GLM-5.3-Flash Z.ai 69±1 local ~103GB 1M llama.cpp › copy 54 Kimi K2.7 Code Moonshot AI 68±1 local ~359GB 262k llama.cpp › copy 55 INTELLECT-3 Prime Intellect 67±2 † local ~48GB llama.cpp › copy 57 Inkling Small (266B-A12B) Thinking Machines 67±1 local ~86GB 1M llama.cpp › copy 58 GLM-4.7 Z.ai 67±2 local ~58GB 205k llama.cpp › copy 60 LongCat 2.0 (1.6T-A48B) Meituan 67±6 * † local ~2155GB 1M how to run › copy 61 DeepSeek V4 Pro (Preview, Apr 2026) DeepSeek 67±1 local ~604GB llama.cpp › copy 62 Solar Open 2 (250B-A15B) Upstage 67±3 * † local ~527GB 1M how to run › copy 66 Qwen3-Coder-Next (80B-A3B) Alibaba 66±5 * local ~27GB 262k llama.cpp › copy 70 Qwen3 235B-A22B Thinking 2507 Alibaba 66±1 local ~76GB 131k llama.cpp › copy 71 EXAONE 4.5 33B LG AI Research 65±4 * † local ~24GB 262k llama.cpp › copy 72 Kimi K2.5 Moonshot AI 65±2 local ~160GB 262k llama.cpp › copy 73 GLM-5 Z.ai 65±1 local ~119GB 205k llama.cpp › copy 74 Granite 4.2 30B IBM 65±3 * † local ~19GB 131k LM Studio › copy 75 DeepSeek V3.2 Exp DeepSeek 65±2 local ~726GB how to run › copy 77 GLM-5.1 Z.ai 65±1 local ~119GB 200k llama.cpp › copy 80 DeepSeek V3.2 DeepSeek 65±2 local ~110GB 128k llama.cpp › copy 81 Qwen3.5 122B-A10B Alibaba 64±2 † local ~75GB 262k LM Studio › copy 82 Step 3.5 Flash (196B-A11B) StepFun 64±1 local ~118GB 256k LM Studio › copy 84 Qwen3.6 27B Alibaba 64±1 local ~18GB 262k LM Studio › copy 88 Qwen3.5 397B-A17B Alibaba 64±1 local ~65GB 262k llama.cpp › copy +1 89 MiMo-V2-Flash (309B-A15B) Xiaomi 64±2 local ~108GB llama.cpp › copy 91 GLM-4.6 Z.ai 64±2 † local ~58GB 205k llama.cpp › copy 92 Spark-X2.5 4B iFLYTEK (SparkLLM) 64±2 † local ~3.9GB 1M LM Studio › copy 93 GigaChat 3.5 Ultra Reasoning (432B-A28B) Sber 64±2 † local ~461GB 262k how to run › copy 94 Inkling (975B-A41B) Thinking Machines 64±1 local ~156GB 1M llama.cpp › copy 96 Qwen3 235B-A22B Instruct 2507 Alibaba 63±1 local ~76GB 262k llama.cpp › copy 97 DeepSeek V3.1 DeepSeek 63±1 local ~110GB 131k llama.cpp › copy 98 GLM-4.5-Air Z.ai 63±2 local ~35GB 131k llama.cpp › copy 99 Qwen3.5 27B Alibaba 63±1 local ~18GB 262k LM Studio › copy 101 Nemotron Cascade 2 30B-A3B NVIDIA 63±2 † local ~20GB 256k LM Studio › copy 102 K-EXAONE (236B-A23B) LG AI Research 63±2 † local ~152GB 262k llama.cpp › copy 104 Gemma 4 26B-A4B Google 63±2 local ~18GB 262k LM Studio › copy 105 Qwen3-Next 80B-A3B Instruct Alibaba 63±3 † local ~27GB 131k llama.cpp › copy 106 DeepSeek R1 0528 DeepSeek 63±1 local ~110GB 164k llama.cpp › copy 107 Kimi K2 Thinking Moonshot AI 63±3 * local ~160GB 262k llama.cpp › copy 108 Step 3.7 Flash (198B-A11B) StepFun 63±1 local ~104GB 256k llama.cpp › copy 110 gpt-oss-120b OpenAI 62±1 local ~67GB 131k llama.cpp › copy 111 Command A+ (05-2026) Cohere 62±2 local ~71GB 128k llama.cpp › copy 112 Qwen3-VL 32B Thinking Alibaba 62±2 local ~22GB 256k LM Studio › copy 115 Nemotron 3 Ultra 550B-A55B NVIDIA 62±2 local ~89GB 1M llama.cpp › copy 116 K-EXAONE 2.0 (750B-A37B) LG AI Research 62±2 † local ~791GB 262k how to run › copy 117 Nemotron-Labs-3 Puzzle 75B-A9B NVIDIA 62±3 * † local ~35GB llama.cpp › copy 118 GLM-4.5 Z.ai 62±1 local ~58GB 131k llama.cpp › copy 119 Qwen3-VL 235B-A22B Instruct Alibaba 62±2 † local ~134GB 256k LM Studio › copy 120 Olmo 3.1 32B Think Ai2 62±2 † local ~21GB 66k LM Studio › copy 123 Gemma 4 31B Google 61±2 local ~21GB 262k LM Studio › copy 124 Qwen3-VL 235B-A22B Thinking Alibaba 61±2 † local ~111GB 256k LM Studio › copy 125 Nemotron 3 Super 120B-A12B NVIDIA 61±1 local ~73GB 262k LM Studio › copy 128 Nemotron-Cascade-14B-Thinking NVIDIA 61±2 * † local ~10GB LM Studio › copy 129 A.X K1 (519B-A33B) SK Telecom 61±4 * † local ~1091GB how to run › copy 130 Granite 4.2 8B IBM 61±3 * † local ~6.6GB 131k LM Studio › copy 131 Falcon-H1R 7B TII 61±2 † local ~6GB 262k LM Studio › copy 132 Qwen3-Next 80B-A3B Thinking Alibaba 61±2 † local ~27GB 131k llama.cpp › copy 133 Qwen3-VL 30B-A3B Thinking Alibaba 61±3 * † local ~21GB 256k LM Studio › copy 134 Qwen3.5 35B-A3B Alibaba 61±1 local ~23GB 262k LM Studio › copy 135 Qwen3.6 35B-A3B Alibaba 61±1 local ~23GB 262k LM Studio › copy 136 LongCat-Flash-Chat (560B-A27B) Meituan 61±2 † local ~333GB 128k LM Studio › copy 139 Nemotron-Cascade-8B NVIDIA 60±2 * † local ~6.3GB LM Studio › copy 141 Kimi K2 Instruct (0711) Moonshot AI 60±2 * local ~160GB 131k llama.cpp › copy 143 Olmo 3 32B Think Ai2 60±1 † local ~21GB 66k LM Studio › copy 146 gpt-oss-20b OpenAI 60±1 local ~14GB 131k llama.cpp › copy 147 MiniCPM5 2B OpenBMB 60±2 † local ~3GB 131k LM Studio › copy 148 MiniMax-M3 MiniMax 60±2 * local ~137GB 1M llama.cpp › copy 149 Seed-OSS 36B Instruct ByteDance Seed 60±2 local ~23GB 524k LM Studio › copy 150 Qwen3 30B-A3B Instruct 2507 Alibaba 59±1 local ~20GB 262k LM Studio › copy 153 Nemotron 3 Nano 30B-A3B NVIDIA 59±2 † local ~20GB 262k LM Studio › copy 155 Qwen3.5 9B Alibaba 59±1 local ~7.4GB 262k llama.cpp › copy 156 Qwen3 14B Alibaba 59±2 * local ~10GB 33k LM Studio › copy 157 QwQ-32B Alibaba 58±2 local ~21GB 131k LM Studio › copy 158 Qwen3-VL 8B Thinking Alibaba 58±2 local ~6.8GB 256k llama.cpp › copy 160 Qwen3.5 4B Alibaba 58±1 local ~4.3GB 262k llama.cpp › copy 161 DeepSeek R1 (Jan '25) DeepSeek 58±1 local ~110GB 128k llama.cpp › copy 162 EXAONE 4.0 32B LG AI Research 58±2 † local ~20GB 131k LM Studio › copy 163 Granite 4.2 3B IBM 58±1 local ~3.7GB 131k LM Studio › copy 164 DeepSeek R1 0528 Qwen3 8B DeepSeek 58±1 local ~6.3GB 128k LM Studio › copy 165 Qwen3-VL 4B Thinking Alibaba 57±2 * † local ~4.1GB 256k llama.cpp › copy 166 Qwen3 30B-A3B Thinking 2507 Alibaba 57±1 local ~20GB 262k LM Studio › copy 169 Olmo 3 7B Think Ai2 57±1 † local ~5.8GB 66k LM Studio › copy 171 Phi-4-reasoning-plus 14B Microsoft 57±1 local ~10GB LM Studio › copy 172 Qwen3-VL 32B Instruct Alibaba 57±2 * † local ~22GB 256k LM Studio › copy 173 GLM-4.7-Flash Z.ai 57±2 * local ~19GB 200k LM Studio › copy 174 Gemma 4 12B Google 57±2 local ~8.5GB 262k LM Studio › copy 176 MiMo-V2.5 (310B-A15B) Xiaomi 57±2 * local ~100GB 1M llama.cpp › copy 177 Nemotron Nano 9B v2 NVIDIA 57±1 local ~7.6GB 131k llama.cpp › copy 179 Qwen3-VL 30B-A3B Instruct Alibaba 56±2 * † local ~21GB 256k LM Studio › copy 180 MiniMax-M2.7 MiniMax 56±2 * local ~75GB 205k llama.cpp › copy 181 DeepSeek R1 Distill Llama 70B DeepSeek 56±1 local ~30GB 131k llama.cpp › copy 184 OpenThinker3 7B Open Thoughts 56±1 local ~6GB 33k LM Studio › copy 185 OpenThinker2 32B Open Thoughts 56±2 † local ~21GB 33k LM Studio › copy 186 Qwen3 32B Alibaba 56±1 local ~20GB 131k LM Studio › copy 187 Qwen3 8B Alibaba 55±1 local ~6.1GB 33k LM Studio › copy 188 DeepSeek R1 Distill Qwen 32B DeepSeek 55±1 local ~21GB 131k LM Studio › copy 189 DeepSeek R1 Distill Qwen 14B DeepSeek 55±1 local ~10GB 131k LM Studio › copy 190 Phi-4-reasoning 14B Microsoft 55±2 † local ~10GB 33k LM Studio › copy 192 Mistral Medium 3.5 (128B) Mistral 55±2 * local ~42GB 262k llama.cpp › copy 194 Solar Open 100B (102B-A12B) Upstage 55±2 * † local ~41GB 131k llama.cpp › copy 196 Qwen3 4B Instruct 2507 Alibaba 54±1 local ~3.9GB 262k LM Studio › copy 197 Phi-4-mini-reasoning 3.8B Microsoft 54±2 * † local ~3.8GB 128k LM Studio › copy 198 Magistral Small 1.2 Mistral 54±1 local ~16GB 131k LM Studio › copy 201 OpenThinker 32B Open Thoughts 53±1 † local ~21GB 33k LM Studio › copy 202 Qwen3 4B Thinking 2507 Alibaba 53±2 local ~3.9GB 262k LM Studio › copy 206 QED-Nano LM Provers 52±2 local ~4.1GB 262k llama.cpp › copy 207 Qwen3-VL 8B Instruct Alibaba 52±2 local ~6.8GB 256k llama.cpp › copy 210 DeepSeek R1 Distill Qwen 7B DeepSeek 52±1 local ~6GB 131k LM Studio › copy +1 211 EXAONE Deep 32B LG AI Research 52±2 † local ~22GB 33k Ollama › copy 212 Nemotron 3 Nano 4B NVIDIA 51±1 local ~3.8GB 262k LM Studio › copy 213 s1.1 32B Simple Scaling 51±1 local ~22GB 33k LM Studio › copy 214 Spark-X2.5 1.7B iFLYTEK (SparkLLM) 51±2 † local ~2.7GB 1M llama.cpp › copy 215 DeepSeek V3 0324 DeepSeek 51±1 local ~108GB 164k llama.cpp › copy 217 Qwen3 30B-A3B (Apr 2025, hybrid) Alibaba 50±1 local ~19GB 33k LM Studio › copy 218 Qwen2.5-VL 72B Instruct Alibaba 50±4 * † local ~46GB 33k LM Studio › copy 219 EXAONE Deep 7.8B LG AI Research 50±2 † local ~6.5GB 33k llama.cpp › copy 220 OpenThinker2 7B Open Thoughts 50±2 † local ~6.4GB 33k llama.cpp › copy 221 Llama 4 Maverick (402B-A17B) Meta 50±1 local ~65GB 1M llama.cpp › copy 222 Gemma 3 27B Google 50±1 local ~18GB 131k LM Studio › copy 223 Ministral 3 14B Mistral 50±3 * † local ~9.1GB 262k LM Studio › copy 224 Magistral Small 1.0 (2506) Mistral 50±1 local ~16GB 128k LM Studio › copy 225 Mistral Small 3.2 24B Mistral 50±2 * local ~16GB 128k LM Studio › copy 228 Olmo 3 7B Instruct Ai2 49±1 † local ~5.8GB 66k LM Studio › copy 229 MiniMax-M2.1 MiniMax 49±2 local ~79GB llama.cpp › copy 230 Qwen3 235B-A22B (Apr 2025, hybrid) Alibaba 49±1 local ~76GB 131k llama.cpp › copy 231 Gemma 3 12B Google 49±1 local ~9.1GB 131k LM Studio › copy 232 DeepSeek R1 Distill Llama 8B DeepSeek 49±1 local ~6.2GB 131k LM Studio › copy 233 Magistral Small 1.1 (2507) Mistral 49±2 * † local ~16GB 40k LM Studio › copy 235 Qwen3-VL 4B Instruct Alibaba 49±2 * † local ~4.1GB 256k llama.cpp › copy 236 GigaChat 3.5 Ultra Instruct (432B-A28B) Sber 49±2 † local ~457GB 262k how to run › copy 238 Command A Reasoning (08-2025) Cohere 49±4 * † local ~34GB llama.cpp › copy 239 LFM2.5 1.2B Thinking Liquid AI 49±3 † local ~2.2GB 33k LM Studio › copy 240 Ministral 3 8B Mistral 48±2 local ~6.5GB 262k LM Studio › copy 242 LFM2.5 8B-A1B Liquid AI 48±1 local ~6.5GB 128k LM Studio › copy 243 Qwen2.5-Math 72B Instruct Alibaba 48±2 † local ~44GB 4k LM Studio › copy 244 Qwen3 4B Alibaba 48±1 local ~3.7GB 33k LM Studio › copy 245 LFM2 24B-A2B Liquid AI 48±2 local ~16GB 33k LM Studio › copy 247 Pixtral Large (2411) Mistral 48±7 * † local ~51GB llama.cpp › copy 248 DeepSeek V3 (Dec '24) DeepSeek 48±1 local ~337GB 131k llama.cpp › copy 249 Phi-4 14B Microsoft 48±1 local ~10GB 16k LM Studio › copy 251 Qwen3-VL 2B Thinking Alibaba 48±2 * † local ~2.7GB 256k llama.cpp › copy 254 Qwen2.5 72B Instruct Alibaba 47±1 local ~30GB 33k llama.cpp › copy 255 Llama 4 Scout (109B-A17B) Meta 47±1 * local ~36GB 328k llama.cpp › copy 257 Qwen2-Math 72B Instruct Alibaba 47±2 † local ~32GB 4k llama.cpp › copy 258 Granite 4.0 H Small (32B-A9B) IBM 47±1 local ~21GB 131k LM Studio › copy 259 DeepSeek-V2.5 (1210) DeepSeek 47±4 * † local ~82GB llama.cpp › copy 261 Tulu 2 70B Allen Institute for AI 46±5 * † local ~28GB llama.cpp › copy 262 OpenThinker 7B Open Thoughts 46±2 † local ~6GB 33k LM Studio › copy 264 EXAONE Deep 2.4B LG AI Research 46±2 † local ~3.2GB 33k Ollama › copy 265 Qwen2.5-Math 7B Instruct Alibaba 46±2 † local ~6GB 4k LM Studio › copy 266 Qwen2.5 32B Instruct Alibaba 46±1 local ~21GB 131k LM Studio › copy 267 Llama 3.1 Nemotron 51B Instruct NVIDIA 46±5 * † local ~34GB llama.cpp › copy 268 LFM2.5 2.6B Liquid AI 46±1 local ~3.2GB 131k LM Studio › copy 271 Tulu 2 DPO 70B Ai2 45±5 * † local ~32GB llama.cpp › copy 273 Qwen2.5 14B Instruct Alibaba 45±2 local ~10GB 131k LM Studio › copy 274 Granite 4.0 H Tiny (7B-A1B) IBM 45±2 † local ~5.6GB 131k LM Studio › copy 275 DeepSeek R1 Distill Qwen 1.5B DeepSeek 45±1 local ~2.6GB 131k LM Studio › copy 281 Llama 3.1 Tulu 3 70B Ai2 45±3 * † local ~43GB 131k LM Studio › copy 282 Qwen2.5-VL 7B Instruct Alibaba 45±3 * † local ~6.4GB llama.cpp › copy 283 Command A (03-2025) Cohere 44±3 * † local ~37GB 256k llama.cpp › copy 284 Gemma 4 E4B Google 44±1 local ~6.9GB 131k LM Studio › copy 285 Mistral Large 2 (2411) Mistral 44±1 * local ~78GB 128k llama.cpp › copy 286 Qwen2-VL 72B Instruct Alibaba 44±3 local ~34GB llama.cpp › copy 287 Llama 3.1 405B Instruct Meta 44±1 local ~160GB 128k llama.cpp › copy 289 Jamba 1.5 Large AI21 Labs 44±4 * † local ~839GB how to run › copy 290 Mixtral 8x7B v0.1 (base) Mistral AI 44±2 * † local ~29GB 32k llama.cpp › copy 291 EXAONE 4.0 1.2B LG AI Research 44±2 * † local ~2.2GB 66k LM Studio › copy 292 Qwen2.5 7B Instruct Alibaba 44±1 local ~6GB 131k LM Studio › copy 293 SmolLM3 3B Hugging Face 44±2 local ~3.3GB 66k LM Studio › copy 294 Gemma 3 4B Google 43±2 local ~3.8GB 131k LM Studio › copy 295 Mistral Small 3.1 24B (2503) Mistral 43±1 local ~17GB 128k llama.cpp › copy 297 Granite 3.3 8B Instruct IBM 43±1 local ~6.3GB 128k LM Studio › copy 299 Mistral Large 2 (2407) Mistral 43±1 local ~45GB 128k llama.cpp › copy 300 InternVL2-26B Shanghai AI Laboratory (OpenGVLab) 43±7 * † local ~55GB how to run › copy 301 DeepSeek-V2 Chat (0628) DeepSeek 43±3 * local ~66GB llama.cpp › copy 302 InternVL2-4B Shanghai AI Laboratory (OpenGVLab) 43±5 * local ~10GB how to run › copy 303 DeepSeek-Coder-V2 Instruct (0614) DeepSeek 43±3 * local ~92GB llama.cpp › copy 304 Xwin-LM 70B v0.1 Xwin-LM 43±5 * local ~28GB llama.cpp › copy 305 Qwen3 1.7B Alibaba 43±1 local ~2.6GB 33k LM Studio › copy +1 306 Mistral Small 3 24B (2501) Mistral 43±1 * local ~16GB 32k LM Studio › copy 307 Granite 4.0 H Micro 3B IBM 43±2 † local ~3.4GB 131k LM Studio › copy 308 Llama 3.1 Tulu 3 70B DPO Ai2 43±1 local ~22GB 131k llama.cpp › copy 309 Llama 3.3 70B Instruct Meta 42±1 local ~24GB 128k llama.cpp › copy 310 Ministral 3 3B Mistral 42±1 local ~3.5GB 262k LM Studio › copy 311 DeepSeek-V2.5 DeepSeek 42±1 local ~151GB 164k Ollama › copy 312 Qwen2-Math 7B Instruct Alibaba 42±2 † local ~6GB 4k LM Studio › copy 313 Granite 4.0 Micro 3B IBM 42±2 local ~3.5GB 131k LM Studio › copy 315 Granite 8B Code Base (4K) IBM 42±2 * † local ~6.6GB llama.cpp › copy 317 Llama 3.2 90B Vision Instruct Meta 42±1 local ~58GB 128k Ollama › copy 318 Qwen2 72B Instruct Alibaba 41±1 local ~33GB 131k llama.cpp › copy 319 Llama 3.1 70B Instruct Meta 41±1 local ~27GB 128k llama.cpp › copy 322 Gemma 4 E2B Google 41±1 local ~5.1GB 131k llama.cpp › copy 323 Qwen2.5-Math 1.5B Instruct Alibaba 41±2 † local ~2.4GB 4k LM Studio › copy 325 Phi-4-mini 3.8B Microsoft 41±2 † local ~3.8GB 128k LM Studio › copy 326 DeepSeek-Coder-V2 Instruct (0724) DeepSeek 41±3 * local ~151GB llama.cpp › copy 327 Phi-2 (2.7B) Microsoft 40±3 * local ~3.2GB 2k LM Studio › copy 328 Molmo 2 8B Ai2 40±7 * † local ~6.8GB llama.cpp › copy 329 Pixtral 12B (2409) Mistral 40±3 * † local ~9.6GB llama.cpp › copy 330 StarCoder2 15B BigCode 40±2 * † local ~11GB 16k LM Studio › copy 331 LFM2 2.6B Liquid AI 40±3 * † local ~3GB 33k LM Studio › copy 332 Qwen3.5 2B Alibaba 40±1 local ~2.9GB 262k llama.cpp › copy 333 Llama 3.1 Nemotron 70B Instruct NVIDIA 40±3 * local ~43GB 128k LM Studio › copy 334 Qwen2.5-VL 3B Instruct Alibaba 40±3 * † local ~3.5GB llama.cpp › copy 335 Molmo 72B (0924) Ai2 40±6 * local ~309GB how to run › copy 336 Qwen3-VL 2B Instruct Alibaba 40±2 * † local ~2.7GB 256k llama.cpp › copy 337 Gemma 7B (base) Google 40±2 local ~7.1GB 8k llama.cpp › copy 338 InternLM2.5 7B Chat Shanghai AI Laboratory (InternLM) 39±3 * † local ~6.1GB LM Studio › copy 339 Qwen2-Math 1.5B Instruct Alibaba 39±2 † local ~2.4GB 4k LM Studio › copy 340 CodeGemma 7B (v1.0) Google 39±2 * local ~7.1GB llama.cpp › copy 341 Code Tulu 2 34B Allen Institute for AI 39±5 * † local ~72GB how to run › copy 342 LFM2 1.2B Liquid AI 39±3 * † local ~2.2GB 33k LM Studio › copy 343 WizardLM 70B V1.0 Microsoft 38±4 * local ~32GB llama.cpp › copy 345 Llama 3.2 11B Vision Instruct Meta 38±3 * local ~8.8GB 128k Ollama › copy 347 LFM2.5 1.2B Instruct Liquid AI 38±2 † local ~2.2GB 33k LM Studio › copy 348 Gemma 2 27B IT Google 38±1 local ~18GB 8k LM Studio › copy 349 WizardLM-2 8x22B Microsoft 38±1 * local ~91GB Ollama › copy 350 Llama 2 70B (base) Meta 38±2 local ~32GB 4k llama.cpp › copy 351 Phi-3-vision-128k-instruct Microsoft 38±6 * † local ~10GB how to run › copy 352 Mathstral 7B v0.1 Mistral AI 38±3 * † local ~5.8GB 32k LM Studio › copy 353 Granite 3.3 2B Instruct IBM 38±2 † local ~3GB 128k LM Studio › copy 355 Phi-3.5-vision-instruct Microsoft 38±7 * † local ~10GB how to run › copy 356 DeepSeek-Coder 33B Base DeepSeek 38±2 * local ~22GB 16k llama.cpp › copy 358 Tulu 2 DPO 13B Allen Institute for AI 37±5 * † local ~9.8GB llama.cpp › copy 359 Llama 3 70B Instruct Meta 37±1 local ~24GB 8k llama.cpp › copy 360 Jamba 1.5 Mini AI21 Labs 37±4 * † local ~34GB llama.cpp › copy 361 Yi 1.5 34B Chat 01.AI 37±1 local ~22GB 4k LM Studio › copy 362 Llama 3.1 Tulu 3 8B Ai2 37±1 local ~6.2GB 131k LM Studio › copy 363 Hermes 2 Theta Llama-3 70B Nous Research 37±1 * local ~47GB 8k llama.cpp › copy 364 Granite 3.2 8B Instruct IBM 37±2 † local ~6.6GB 131k llama.cpp › copy 366 Granite 4.0 1B IBM 37±2 local ~2.4GB 128k LM Studio › copy 367 Qwen2.5 3B Instruct Alibaba 37±1 local ~3.3GB 33k LM Studio › copy 368 OLMo 2 32B Instruct (0325) Ai2 37±1 † local ~21GB LM Studio › copy 369 Llama 3.1 8B Instruct Meta 37±1 local ~6.2GB 128k LM Studio › copy 370 Granite 3.1 8B Instruct IBM 36±2 † local ~6.6GB 128k llama.cpp › copy 371 OLMo 2 13B Instruct (1124) Allen Institute for AI 36±2 * † local ~9.6GB 4k LM Studio › copy 372 Gemma 2 9B IT Google 36±1 local ~7GB 8k LM Studio › copy 373 Tulu 2 13B Allen Institute for AI 36±4 * † local ~9.8GB llama.cpp › copy 374 Granite 3.0 8B Instruct IBM 36±2 * † local ~6.6GB 4k llama.cpp › copy 377 Granite 4.0 H 1B IBM 36±2 local ~2.3GB 131k LM Studio › copy 378 Athene-70B Nexusflow 35±1 * local ~29GB llama.cpp › copy 379 Qwen3 0.6B Alibaba 35±2 local ~1.8GB 33k LM Studio › copy 380 Phi-3 Medium 128K Instruct Microsoft 35±1 * local ~9.8GB 128k LM Studio › copy 381 Code Llama 34B (base) Meta 35±2 * local ~23GB 16k Ollama › copy 382 Qwen2 7B Instruct Alibaba 35±1 local ~6GB 131k LM Studio › copy 384 Ministral 8B (2410) Mistral 34±1 local ~6.2GB 128k LM Studio › copy 385 Code Tulu 2 13B Allen Institute for AI 34±5 * † local ~29GB how to run › copy 387 LFM2 700M Liquid AI 34±3 * † local ~1.9GB 33k LM Studio › copy 390 Phi-3.5-mini-instruct Microsoft 34±1 local ~3.7GB 128k LM Studio › copy 392 Qwen2-VL 7B Instruct Alibaba 34±2 local ~6.4GB LM Studio › copy 393 Code Llama 34B Instruct Meta 33±5 * local ~22GB 16k LM Studio › copy 394 Yi 1.5 9B Chat 01.AI 33±1 * local ~6.8GB 4k LM Studio › copy 396 Granite 3.0 2B Instruct IBM 33±2 * † local ~3.2GB 4k llama.cpp › copy 397 OLMo 2 7B Instruct (1124) Allen Institute for AI 33±2 † local ~5.8GB 4k LM Studio › copy 398 Qwen3.5 0.8B Alibaba 33±4 † local ~2GB 262k llama.cpp › copy 399 Qwen2.5 1.5B Instruct Alibaba 33±2 local ~2.4GB 33k LM Studio › copy 401 Mistral 7B v0.1 (base) Mistral AI 33±1 local ~5.8GB 8k LM Studio › copy 402 Llama 3.2 3B Instruct Meta 32±1 local ~3.4GB 131k LM Studio › copy 404 Mistral NeMo Instruct (2407) Mistral 32±1 * local ~8.7GB 128k LM Studio › copy 405 DBRX Instruct Databricks 32±1 * local ~85GB 32k Ollama › copy 406 Tulu 2 DPO 7B Allen Institute for AI 32±5 * † local ~5.8GB llama.cpp › copy 408 Tulu 2 7B Allen Institute for AI 32±5 * † local ~5.8GB llama.cpp › copy 409 Phi-3-mini-4k-instruct (June 2024 update) Microsoft 32±1 * local ~3.7GB LM Studio › copy 410 Command R Plus (04-2024) Cohere 32±1 * local ~43GB 128k llama.cpp › copy 411 Code Tulu 2 7B Allen Institute for AI 32±5 * † local ~16GB how to run › copy 412 Gemma 3 1B Google 31±1 local ~2.2GB 32k LM Studio › copy 413 DeepSeek-Coder 6.7B Base DeepSeek 31±2 * local ~5.8GB 16k llama.cpp › copy 414 Granite 3.2 2B Instruct IBM 31±2 † local ~3GB 131k LM Studio › copy 415 Granite 3.1 2B Instruct IBM 31±2 † local ~3.1GB 128k llama.cpp › copy 417 Stable Code 3B Stability AI 31±2 * local ~3.3GB 16k llama.cpp › copy 418 Mixtral 8x7B Instruct v0.1 Mistral 30±1 * local ~29GB 32k LM Studio › copy 419 StarCoder2 7B BigCode 30±2 local ~6.1GB 16k LM Studio › copy 420 Granite 3.1 3B-A800M Instruct IBM 30±4 * † local ~3.6GB 131k llama.cpp › copy 421 Zephyr 7B beta Hugging Face 30±5 * local ~6.1GB llama.cpp › copy 422 Granite 3.0 3B-A800M Instruct IBM 30±2 * † local ~3.7GB 4k llama.cpp › copy 423 OLMo 2 1B Instruct (0425) Allen Institute for AI 29±2 * † local ~2.4GB 4k llama.cpp › copy 425 Code Llama 13B (base) Meta 29±2 * local ~9.2GB 16k LM Studio › copy 426 Llama 3 8B Instruct Meta 29±1 local ~6.6GB 8k llama.cpp › copy 427 WizardLM 13B V1.2 Microsoft 29±4 * local ~9.8GB llama.cpp › copy 428 DeepSeek LLM 67B Chat DeepSeek 29±1 * local ~31GB 4k llama.cpp › copy 429 Apertus 8B Instruct (2509) Swiss AI 29±3 * local ~6.3GB 66k LM Studio › copy 430 Code Llama 13B Instruct Meta 28±5 * local ~9.8GB 16k llama.cpp › copy 431 SmolLM2 1.7B Instruct Hugging Face 27±2 local ~2.6GB 8k LM Studio › copy 432 Yi 34B Chat 01.AI 27±1 * local ~23GB 4k llama.cpp › copy 433 LFM2 350M Liquid AI 27±3 * † local ~1.7GB 33k LM Studio › copy 434 Falcon 40B (base) TII 27±3 * local ~27GB 2k Ollama › copy 435 Gemma 2B (base) Google 27±4 * † local ~3.2GB 8k llama.cpp › copy 436 Llama 2 70B Chat Meta 26±1 local ~32GB 4k llama.cpp › copy 437 Llama 2 13B (base) Meta 26±2 local ~9.3GB 4k Ollama › copy 438 StarCoder2 3B BigCode 26±2 * † local ~3.4GB 16k llama.cpp › copy 439 Llama 3.2 1B Instruct Meta 25±1 local ~2.2GB 131k LM Studio › copy 440 Code Llama 7B Instruct Meta 25±5 * local ~5.5GB 16k LM Studio › copy 441 Llama-3-Instruct-8B-SimPO-v0.2 Princeton NLP 25±2 * local ~6.6GB llama.cpp › copy 442 Granite 3.1 1B-A400M Instruct IBM 25±4 * † local ~2.3GB 131k llama.cpp › copy 443 Mistral 7B Instruct v0.3 Mistral 25±1 local ~5.8GB 32k LM Studio › copy 444 Qwen2 0.5B Instruct Alibaba 24±5 * † local ~1.8GB 32k LM Studio › copy 445 Molmo 7B-D (0924) Ai2 24±3 local ~35GB how to run › copy 446 Gemma 2 2B IT Google 24±1 local ~3.1GB 8k LM Studio › copy 447 Command R (03-2024) Cohere 23±1 * local ~24GB 128k llama.cpp › copy 448 Llama 2 7B Chat Meta 23±5 * local ~5.8GB 4k llama.cpp › copy 449 Qwen2-VL 2B Instruct Alibaba 23±3 * † local ~2.8GB 33k LM Studio › copy 450 StarCoderBase 15.5B BigCode 23±2 * † local ~12GB llama.cpp › copy 451 Qwen2 1.5B Instruct Alibaba 23±1 local ~2.4GB 32k LM Studio › copy 452 Granite 3.0 1B-A400M Instruct IBM 23±3 * † local ~2.3GB 4k llama.cpp › copy 453 Code Llama 7B (base) Meta 22±2 local ~5.5GB 16k LM Studio › copy 455 Granite 4.0 H 350M IBM 20±2 † local ~1.7GB 33k LM Studio › copy 456 Granite 4.0 350M IBM 19±2 local ~1.7GB 32k LM Studio › copy 457 Llama 2 13B Chat Meta 19±3 * local ~9.8GB 4k llama.cpp › copy 458 Llama 2 7B (base) Meta 18±2 local ~5.5GB 4k Ollama › copy 460 DeepSeek-Coder 1.3B Base DeepSeek 17±3 * local ~2.4GB 16k Ollama › copy 461 StarCoderBase 7B BigCode 17±2 local ~6GB Ollama › copy 462 StarCoderBase 3B BigCode 13±3 * † local ~3.4GB Ollama › copy — DeepSeek LLM 7B Chat DeepSeek not yet calibrated local ~5.7GB Ollama › copy — DeepSeek-V2 Lite Chat (16B-A2.4B) DeepSeek not yet calibrated local ~11GB 32k LM Studio › copy — ERNIE 4.5 300B-A47B Baidu not yet calibrated local ~97GB 131k llama.cpp › copy — Eurus-2-7B-PRIME PRIME-RL not yet calibrated local ~6.4GB 4k llama.cpp › copy — Eurus-2-7B-SFT PRIME-RL not yet calibrated local ~6.4GB llama.cpp › copy — Flan-T5 Base Google not yet calibrated local ~2.5GB how to run › copy — Flan-T5 Large Google not yet calibrated local ~4.8GB how to run › copy — Flan-T5 Small Google not yet calibrated local ~1.8GB how to run › copy — Flan-T5 XL Google not yet calibrated local ~14GB how to run › copy — Flan-T5 XXL Google not yet calibrated local ~49GB how to run › copy — Gemma 3n E2B IT Google not yet calibrated local ~4.7GB 32k llama.cpp › copy — Gemma 3n E4B Google not yet calibrated local ~6.2GB 32k llama.cpp › copy — GPT-2 XL (1.5B) OpenAI not yet calibrated local ~2.8GB 1k llama.cpp › copy — GPT-Neo 1.3B EleutherAI not yet calibrated local ~7.1GB 2k how to run › copy — GPT-Neo 2.7B EleutherAI not yet calibrated local ~13GB 2k how to run › copy — Granite 4.1 30B IBM not yet calibrated local ~19GB Ollama › copy — Granite 4.1 3B IBM not yet calibrated local ~3.7GB Ollama › copy — Granite 4.1 8B IBM not yet calibrated local ~7GB 131k LM Studio › copy — Hunyuan Hy3 (295B-A21B) Tencent not yet calibrated local ~96GB 256k llama.cpp › copy — Hy3 preview (295B-A21B) Tencent not yet calibrated local ~629GB how to run › copy — Hy4 preview (770B-A49B) Tencent not yet calibrated local ~248GB 1M llama.cpp › copy — K2 Horizon 7B Institute of Foundation Models (MBZUAI) not scored yet local ~7.7GB 524k llama.cpp › copy — Ling 3.0 Flash (124B-A5B) inclusionAI (Ant Group) not yet calibrated local ~75GB 131k LM Studio › copy — Llama 3.1 Nemotron Ultra 253B v1 NVIDIA not yet calibrated local ~82GB 128k llama.cpp › copy — Llama 3.3 Nemotron Super 49B v1 NVIDIA not yet calibrated local ~22GB llama.cpp › copy — MiniMax-M1 (80k) MiniMax not yet calibrated local ~959GB 1M how to run › copy — Ministral 3 14B Reasoning 2512 Mistral not yet calibrated local ~10GB 262k llama.cpp › copy — Ministral 3 3B Reasoning 2512 Mistral not yet calibrated local ~3.5GB 262k LM Studio › copy — Ministral 3 8B Reasoning 2512 Mistral not yet calibrated local ~6.5GB 262k LM Studio › copy — Mistral Small 4 (119B-A6.5B) Mistral not scored yet local ~40GB 262k llama.cpp › copy — Muse Glimmer 30B Meta not scored yet local ~19GB 131k llama.cpp › copy — OLMo 7B Allen Institute for AI not yet calibrated local ~30GB 2k how to run › copy — OLMo 7B Instruct Ai2 not yet calibrated local ~16GB how to run › copy — OpenChat 3.5 OpenChat not yet calibrated local ~6.1GB 8k llama.cpp › copy — OpenHermes 2.5 Mistral 7B Nous Research not yet calibrated local ~6.1GB llama.cpp › copy — Qwen2.5 0.5B Instruct Alibaba not yet calibrated local ~1.8GB 33k LM Studio › copy — SmolLM2 360M Instruct Hugging Face not yet calibrated local ~1.7GB 8k LM Studio › copy — Trinity-Large-Preview Arcee AI not yet calibrated local ~155GB llama.cpp › copy — Trinity-Large-Thinking Arcee AI not yet calibrated local ~141GB 512k llama.cpp › copy 
