Bestes lokales LLM zum Selbstbetrieb (Ollama und vergleichbar). Dieselbe Arena-Open-Source-Rangliste wie die Open-Source-Seite — keine eigene Tabelle, kein VRAM- oder Laptop-Benchmark.
🏆
Auf einen Blick Kimi K3 (Max)
GLM-5.3 (Max)
🎯
Geeignet für Offene Modelle, die Sie selbst betreiben könnten
💡
Score verstehen Score Offizieller Arena-Score
Votes Dialog-Sitzungen
Preis / Kontext Preis / 1M & Kontext
Ausführliche Regeln anzeigen ▾ Der Score ist der offizielle Arena-Score, genau wie auf arena.ai. Höher bedeutet besser. Stimmen basieren auf paarweisen Blind-Duellen. Preis und Kontext stammen von models.dev; ohne Match wird — angezeigt. Die Balkenlänge skaliert vom letzten Platz bis Platz 1. Nicht mit Agent-IPS vergleichen. Dieselben Karten wie die Open-Source-Rangliste. models.dev hat weder VRAM- noch Parameterzahlen, daher ist das keine Laptop-Rangliste. Open Source
Siehe auch Alle Modelle Weitere Modelle # Modell Score Votes Preis Kontext
18 Kimi K3 (Max) 1487.4 35,379 $3 / $15 1049k 26 GLM-5.3 (Max) 1480.7 25,144 $1.4 / $4.4 1M 28 MiMo-V2.6-Pro 1479.5 4,067 $0.435 / $0.87 1049k 34 DeepSeek-V4.1-Flash-20260910 (Max) 1476.0 13,750 $0.15 / $0.6 1M 35 GLM-5.3-Flash 1476.0 38,359 $0.15 / $0.5 1M 37 GLM-5.2 (Max) 1475.4 52,283 $1.4 / $4.4 1M 51 MiMo-V2.5-Pro 1466.5 77,883 $1 / $3 1049k 53 GLM-5.1 1465.4 65,402 $1.4 / $4.4 200k 57 DeepSeek-V4-Pro (High, 20260813) 1464.5 9,797 $1 / $2.5 1049k 59 Kimi K2.6 1460.7 40,026 $0.95 / $4 262k 64 DeepSeek V4 Pro 1457.9 57,588 $0.66 / $1.98 1M 65 GLM-5 1457.4 29,267 $1 / $3.2 205k 73 DeepSeek-V4-Pro Preview (High) 1454.9 54,896 $1 / $2.5 1049k 74 Hy3 1453.8 13,220 $0.132 / $0.528 262k 76 Gemma 4 31B 1452.3 6,137 $0.144 / $0.42 262k 78 Kimi K2.5 (Thinking) 1449.9 74,744 $0.627 / $3.3 262k 81 MiMo-V2.6-Flash 1448.1 13,730 $0.14 / $0.28 1049k 84 Step 5 Preview 1447.7 9,792 $1 / $2.7 1024k 92 Qwen3.5 397B-A17B 1441.7 96,206 $0.6 / $3.6 262k 94 Inkling 1441.6 43,705 $1 / $4.05 1049k 95 GLM-4.7 1441.3 12,245 $0.6 / $2.2 205k 97 MiniMax-M3 1439.8 67,030 $0.3 / $1.2 1049k 98 DeepSeek-V4-Flash Preview (High) 1438.7 51,781 $0.39 / $1.56 1M 100 Gemma 4 26B A4B 1437.5 6,092 $0.05 / $0.29 262k 102 Qwen3.8 27B 1437.1 26,238 $0.5 / $3 1M 104 Beam 1.0 501B A23B 1435.9 4,237 $0.25 / $1 1049k 105 DeepSeek V4 Flash 1435.7 51,982 $0.22 / $0.66 1M 109 MiMo-V2.5 1434.0 47,364 $0.4 / $2 1049k 114 Kimi K2.5 (Instant) 1430.7 8,371 $0.627 / $3.3 262k 117 Kimi K2 Thinking Turbo 1430.0 63,348 $0.575 / $2.3 262k 119 Mistral Medium 3.5 1426.4 11,791 $1.5 / $7.5 262k 122 NVIDIA-Nemotron-3-Ultra-550B-A55B (NVFP4) 1426.1 11,557 $0.625 / $3.125 256k 123 DeepSeek-V3.2-Exp (Thinking) 1425.2 8,994 $0.28 / $0.42 164k 124 DeepSeek V3.2 1424.8 48,090 $0.293 / $0.4395 128k 125 Muse Glimmer 1424.7 3,889 — — 127 GLM-4.6 1424.3 35,818 $0.6 / $2.2 205k 129 DeepSeek-V3.2 (Thinking) 1422.4 42,042 $0.62 / $1.85 164k 130 Qwen3-235B-A22B-Instruct (2507) 1422.2 98,096 $0.13 / $0.6 262k 132 DeepSeek-R1 (0528) 1421.4 18,170 $3 / $7 164k 133 Kimi K2 Preview (0905) 1418.7 11,586 $0.57 / $2.3 256k 136 Kimi K2 Preview (0711) 1417.8 27,136 $0.57 / $2.3 256k 137 DeepSeek V3.1 1417.0 14,607 $0.27 / $1 164k 138 Qwen3.5 122B A10B (Alibaba Cloud) 1416.6 29,716 $0.4 / $3.2 262k 140 DeepSeek-V3.1 (Thinking) 1415.8 11,454 $0.2 / $0.8 75k 141 MiniMax-M2.7 1415.0 87,731 $0.3 / $1.2 205k 145 Mistral Large 3 1414.3 86,928 $0.5 / $1.5 262k 147 Qwen3 VL 235B A22B Instruct 1413.3 11,312 $0.2 / $0.88 262k 149 NVIDIA-Nemotron-3.5-Super-VL-120B-A12B (NVFP4) 1412.8 4,176 $0.15 / $0.5 1049k 151 GLM-4.5 1410.9 23,791 $0.6 / $2.2 131k 154 Hy3 preview 1409.0 6,875 $0.17 / $0.566661 256k 155 Qwen3.5 27B (Alibaba Cloud) 1408.7 28,661 $0.3 / $2.4 262k 158 Inkling Small 1404.5 35,359 $0.45 / $1.2 524k 161 Qwen3-235B-A22B (Non-Thinking) 1402.1 37,485 $0.0875 / $0.35 262k 165 Qwen3-235B-A22B-Thinking (2507) 1399.8 8,821 $0.3 / $2.9 262k 166 Qwen3 Next 80B A3B Instruct 1398.5 22,755 $0.15 / $1.2 262k 167 DeepSeek-R1 1398.1 18,524 $1.35 / $5.4 128k 169 DeepSeek-V3 (0324) 1395.2 44,974 $0.89 / $0.89 64k 170 Qwen3 VL 235B A22B Thinking 1394.8 7,855 $0.98 / $3.95 131k 172 Qwen3.5 35B A3B (Alibaba Cloud) 1393.9 30,443 $0.25 / $2 262k 174 Step 3.5 Flash 1393.0 60,329 $0.1 / $0.3 256k 175 MiMo-V2-Flash (Non-Thinking) 1391.6 48,069 $0.1 / $0.3 256k 178 MiniMax M2.5 1390.7 42,791 $0.3 / $1.2 205k 182 Qwen3 Coder 1387.5 25,192 $1.058 / $5.15775 262k 186 MiMo-V2-Flash (Thinking) 1386.3 11,140 $0.1 / $0.3 256k 187 MiniMax-M2.1 Preview 1384.6 17,230 $0.27 / $1.1 205k 190 Qwen3-30B-A3B-Instruct (2507) 1382.2 23,306 — — 192 Trinity Large Preview 1378.6 31,246 $0 / $0 131k 193 GLM-4.6V 1378.2 2,831 $0.3 / $0.9 128k 194 Qwen3-235B-A22B 1374.9 25,949 $0.29 / $2.86 128k 198 GLM-4.5-Air 1373.0 30,646 $0.2 / $1.1 131k 200 Qwen3 Next 80B A3B Thinking 1369.3 13,473 $0.15 / $1.2 262k 201 Trinity Large Thinking 1367.4 30,307 — — 202 Gemma 3 27B IT 1365.3 46,665 $0.23 / $0.38 131k 203 GLM-4.7-Flash 1364.7 12,013 $0 / $0 200k 208 Nemotron 3 Super 120B A12B 1360.1 7,869 $0.3 / $2.4 262k 210 DeepSeek V3 1358.4 21,770 $0.27 / $1.1 164k 212 Mistral Small 3.2 (2506) 1356.0 17,454 $0.075 / $0.2 128k 213 INTELLECT-3 1355.7 5,432 — — 214 Command A (03-2025) 1353.7 55,758 $2.5 / $10 256k 216 NVIDIA-Nemotron-3.5-Lightning-30B-A3B (NVFP4) 1353.0 22,498 $0.07 / $0.2 262k 217 GLM-4.5V 1352.2 4,821 $0.6 / $1.8 64k 218 GPT OSS 120B 1351.8 30,207 $0.15 / $0.6 131k 220 Step-3 1349.1 6,295 — — 224 Llama 3.1 Nemotron Ultra 253B v1 1347.6 2,549 $0 / $0 128k 225 Qwen3 32B 1346.8 3,926 $0.36 / $0.87 41k 229 MiniMax M2 1343.6 6,855 $0.3 / $1.2 204k 230 inclusionAI/Ling-flash-2.0 1343.4 6,953 $0.14 / $0.57 131k 232 Llama 3.3 Nemotron Super 49B v1.5 1343.0 3,328 $0.4 / $0.4 131k 235 Gemma 3 12B IT 1341.8 3,829 $0.09 / $0.29 131k 237 Granite 4.2 30B 1340.0 3,407 — — 242 QwQ 32B 1335.6 25,174 $0.4 / $0.4 33k 244 Llama 3.1 405B Instruct (BF16) 1335.5 41,375 — — 246 Llama 3.1 405B Instruct (FP8) 1333.6 59,656 — — 248 Olmo 3.1 32B Instruct 1329.4 11,805 — — 250 Llama 3.3 Nemotron Super 49B v1 1327.7 2,218 $0.15 / $0.15 128k 251 Llama 4 Maverick 17B 128E Instruct 1326.9 39,444 $0.35 / $1.15 524k 252 Qwen3-30B-A3B 1326.7 26,155 $0.11 / $1.08 128k 257 DeepSeek-V2.5 (1210) 1323.4 6,795 — — 258 Ring-flash-2.0 1322.4 6,987 — — 261 Molmo 2 8B 1321.9 823 — — 262 Llama 4 Scout 17B 16E Instruct 1321.5 29,830 $0.2 / $0.78 128k 266 Llama 3.3 70B Instruct 1318.0 54,493 $0.72 / $0.72 128k 268 Gemma 3n E4B IT 1317.6 22,142 — — 269 GPT OSS 20B 1317.4 10,417 $0.07 / $0.25 131k 271 Mistral Large 2 (2407) 1314.4 45,459 $2 / $6 128k 272 Athene V2 Chat 1314.1 24,739 — — 273 NVIDIA-Nemotron-3-Nano-30B-A3B (BF16) 1313.7 15,765 $0.075 / $0.3 128k 279 DeepSeek-V2.5 1307.2 24,572 — — 280 Athene 70B (0725) 1306.6 19,621 — — 281 Olmo 3 32B Think 1306.2 5,825 — — 284 Magistral Medium (2506) 1304.7 11,509 $2 / $5 262k 285 Granite 4.1 8B 1304.2 4,177 — — 286 Gemma 3 4B IT 1303.3 4,171 $0.04 / $0.08 131k 288 Mistral Small 3.1 24B Instruct (2503) 1302.7 32,721 $0.351 / $0.555 128k 289 Llama 3.1 Nemotron 70B Instruct 1298.7 7,140 $0 / $0 128k 291 Llama 3.1 70B Instruct 1293.4 55,240 $0.72 / $0.72 128k 292 Granite 4.2 3B 1290.3 3,260 — — 294 Granite 4.2 8B 1289.8 3,259 $0.06 / $0.25 131k 295 Gemma 2 27B IT 1289.5 75,754 $0.65 / $0.65 8k 296 Jamba 1.5 Large 1289.5 8,662 — — 299 Granite-4.0-H-Small 1287.6 5,628 $0.0636 / $0.265 131k 300 Llama 3.1 Nemotron 51B Instruct 1286.7 3,749 — — 302 Olmo 3.1 32B Think 1286.4 8,299 — — 303 Llama 3.1 Tulu 3 70B 1286.1 2,846 — — 305 Gemma 2 9B IT SimPO 1280.4 10,072 — — 306 Nemotron-4 340B Instruct 1277.2 19,659 — — 307 Llama 3 70B Instruct 1276.6 156,876 — — 308 Command R+ (08-2024) 1276.4 9,866 $0.15 / $0.6 128k 310 Mistral Small 24B Instruct (2501) 1274.4 14,681 — — 313 Qwen2.5 Coder 32B Instruct 1270.6 5,432 $0.66 / $1 33k 314 Aya Expanse 32B 1267.3 27,124 $0.5 / $1.5 128k 316 DeepSeek-Coder-V2 1265.4 15,147 — — 317 Qwen2-72B-Instruct 1261.9 37,325 — — 318 Command R+ 1261.8 77,554 $0.15 / $0.6 128k 323 OLMo 2 32B Instruct (0325) 1251.7 3,334 — — 324 Command R (08-2024) 1250.5 10,140 $0.15 / $0.6 128k 327 Jamba 1.5 Mini 1239.8 8,858 — — 328 Ministral 8B (2410) 1237.6 4,781 $0.15 / $0.15 262k 330 Qwen1.5-110B-Chat 1234.2 26,195 — — 332 Qwen1.5-72B-Chat 1233.6 39,302 — — 334 Mixtral 8x22B Instruct v0.1 1229.7 51,416 $2 / $6 66k 335 Command R 1226.9 54,036 $0.15 / $0.6 128k 338 Llama 3 8B Instruct 1223.7 104,642 — — 340 Aya Expanse 8B 1223.1 9,818 — — 341 Mistral Medium 1222.8 34,550 $0.44 / $2.2 131k 342 Llama 3.1 Tulu 3 8B 1220.5 2,896 — — 343 Zephyr ORPO 141B A35B v0.1 1213.1 4,652 — — 344 Yi 1.5 34B Chat 1212.9 24,146 — — 345 Llama 3.1 8B Instruct 1211.2 49,605 $0.22 / $0.22 128k 346 Granite 3.1 8B Instruct 1208.6 3,090 — — 348 Qwen1.5-32B-Chat 1204.1 21,741 — — 349 Gemma 2 2B IT 1200.1 46,616 — — 350 Phi-3 Medium 4K Instruct 1198.0 25,055 — — 351 Mixtral 8x7B Instruct v0.1 1197.2 73,503 $0.488 / $0.758 32k 352 DBRX Instruct Preview 1195.6 32,191 — — 353 Qwen1.5-14B-Chat 1191.1 17,839 — — 354 InternLM2.5 20B Chat 1190.8 9,901 — — 355 DeepSeek LLM 67B Chat 1185.0 4,932 — — 356 WizardLM 70B 1184.8 8,214 — — 357 Yi 34B Chat 1183.8 15,483 — — 358 Granite 3.0 8B Instruct 1183.2 6,638 — — 359 OpenChat 3.5 1183.1 7,968 — — 360 OpenChat 3.5 (0106) 1182.8 12,637 — — 361 Gemma 1.1 7B IT 1182.6 23,893 — — 362 Snowflake Arctic Instruct 1180.1 32,832 — — 363 Granite 3.1 2B Instruct 1178.9 3,188 — — 364 Tulu 2 DPO 70B 1177.7 6,535 — — 365 OpenHermes 2.5 Mistral 7B 1176.0 5,006 — — 366 Vicuna 33B 1173.0 22,479 — — 367 Phi-3 Small 8K Instruct 1171.1 17,766 — — 368 Starling LM 7B Beta 1171.0 16,056 — — 369 Llama 2 70B Chat 1170.7 38,492 — — 370 Starling LM 7B Alpha 1167.5 10,224 — — 371 Llama 3.2 3B Instruct 1166.8 7,936 $0.02 / $0.02 16k 372 Nous Hermes 2 Mixtral 8x7B DPO 1164.5 3,777 — — 373 Granite 3.0 2B Instruct 1156.7 6,837 — — 374 Llama 2 70B SteerLM Chat 1154.8 3,585 — — 375 QwQ 32B 1154.3 3,231 $0.4 / $0.4 33k 376 SOLAR 10.7B Instruct v1.0 1152.5 4,155 — — 377 Dolphin 2.2.1 Mistral 7B 1152.3 1,679 — — 378 MPT 30B Chat 1150.8 2,572 — — 379 WizardLM 13B 1149.6 7,044 — — 381 Falcon 180B Chat 1148.2 1,295 — — 382 Qwen1.5-7B-Chat 1143.9 4,737 — — 383 Phi-3 Mini 4K Instruct (June 2024) 1143.2 12,297 — — 384 Vicuna 13B 1141.6 19,367 — — 385 Llama 2 13B Chat 1141.5 19,174 — — 386 Qwen-14B-Chat 1139.4 4,964 — — 388 Gemma 7B IT 1137.8 8,925 — — 389 Code Llama 34B Instruct 1136.9 7,366 — — 390 Zephyr 7B Beta 1130.9 11,118 — — 391 Phi-3 Mini 128K Instruct 1129.9 20,685 — — 392 Phi-3 Mini 4K Instruct 1128.2 20,118 — — 393 Guanaco 33B 1127.6 2,921 — — 394 Zephyr 7B Alpha 1127.0 1,785 — — 395 StripedHyena Nous 7B 1121.7 5,182 — — 396 Code Llama 70B Instruct 1119.2 1,143 — — 397 Gemma 1.1 2B IT 1116.7 10,854 — — 398 Vicuna 7B 1115.4 6,923 — — 399 SmolLM2 1.7B Instruct 1114.8 2,199 — — 400 Llama 3.2 1B Instruct 1111.2 8,045 $0.027 / $0.201 60k 401 Mistral 7B Instruct 1110.4 8,977 — — 402 Llama 2 7B Chat 1108.0 14,148 — — 403 Gemma 2B IT 1093.7 4,780 — — 404 Qwen1.5-4B-Chat 1091.2 7,597 — — 405 OLMo 7B Instruct 1073.7 6,328 — — 406 Koala 13B 1071.0 6,965 — — 407 Alpaca 13B 1070.3 5,745 — — 408 GPT4All 13B Snoozy 1067.8 1,743 — — 409 MPT 7B Chat 1063.5 3,924 — — 410 ChatGLM3 6B 1056.6 4,658 — — 411 RWKV-4 Raven 14B 1042.4 4,845 — — 412 ChatGLM2 6B 1024.8 2,658 — — 413 OpenAssistant Pythia 12B 1023.6 6,310 — — 414 ChatGLM 6B 995.9 4,914 — — 415 FastChat-T5 3B 992.6 4,203 — — 416 Dolly 2.0 12B 982.5 3,412 — — 417 LLaMA 13B 975.2 2,391 — — 418 StableLM Tuned Alpha 7B 953.3 3,287 — — 
Best LLM Models