Alibaba Cloud / Qwen Team model product
Qwen2-72B-Instruct is an instruction-tuned language model with 72 billion parameters, supporting a context length of up to 131,072 tokens. It's part of the new Qwen2 series, which has surpassed most open-source models and demonstrates competitiveness against proprietary models across various benchmarks.
Updated Aug 10, 2026. Default version: Qwen2 72B Instruct
Structured fields from the published default version.
This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.
Published benchmark records for the scored version Qwen2 72B Instruct.
| TheoremQA | 0.4 | 1 | 6 | 100.0% | C | |
| CMMLU | 0.9 | 2 | 6 | 80.0% | C | |
| EvalPlus | 0.8 | 2 | 4 | 66.7% | C | |
| Winogrande | 0.9 | 4 | 22 | 85.7% | C | |
| BBH | 0.8 | 7 | 12 | 45.5% | C | |
| HellaSwag | 0.9 | 8 | 27 | 73.1% | C | |
| MultiPL-E | 0.7 | 10 | 13 | 25.0% | C | |
| MBPP | 0.8 | 13 | 37 | 66.7% | C | |
| TruthfulQA | 0.5 | 13 | 18 | 29.4% | C | |
| C-Eval | 0.8 | 14 | 18 | 23.5% | C | |
| ARC-C | 0.7 | 24 | 34 | 30.3% | C | |
| GSM8k | 0.9 | 25 | 48 | 48.9% | C | |
| HumanEval | 0.9 | 36 | 76 | 53.3% | C | |
| MATH | 0.6 | 48 | 71 | 32.9% | C | |
| MMLU | 0.8 | 49 | 100 | 51.5% | C | |
| MMLU-Pro | 0.6 | 101 | 129 | 21.9% | C | |
| GPQA | 0.4 | 198 | 233 | 15.1% | C |
Preference and agent-evaluation signals from published Arena datasets.
| text style control | japanese | 213 | 1175.7 | 901 | N/A | |
| text | japanese | 218 | 1111.4 | 901 | N/A | |
| text style control | french | 232 | 1250.5 | 596 | N/A | |
| text | french | 233 | 1169.0 | 596 | N/A | |
| text style control | spanish | 233 | 1245.6 | 596 | N/A | |
| text | korean | 234 | 1082.8 | 1,167 | N/A | |
| text style control | korean | 234 | 1171.6 | 1,167 | N/A | |
| text | spanish | 237 | 1168.2 | 596 | N/A | |
| text | chinese | 246 | 1239.0 | 4,078 | N/A | |
| text style control | german | 246 | 1220.9 | 1,340 | N/A | |
| text style control | chinese | 247 | 1298.4 | 4,078 | N/A | |
| text | german | 251 | 1150.7 | 1,340 | N/A | |
| text style control | math | 251 | 1273.2 | 4,835 | N/A | |
| text style control | industry mathematical | 252 | 1275.1 | 4,235 | N/A | |
| text | math | 259 | 1234.8 | 4,835 | N/A | |
| text | industry mathematical | 260 | 1230.7 | 4,235 | N/A | |
| text style control | expert | 263 | 1258.6 | 1,763 | N/A | |
| text | expert | 270 | 1170.4 | 1,763 | N/A | |
| text style control | industry medicine and healthcare | 270 | 1274.3 | 2,067 | N/A | |
| text style control | hard prompts english | 277 | 1284.1 | 6,483 | N/A | |
| text | hard prompts english | 280 | 1207.6 | 6,483 | N/A | |
| text | industry medicine and healthcare | 280 | 1174.9 | 2,067 | N/A | |
| text style control | hard prompts | 280 | 1272.8 | 10,780 | N/A | |
| text | multi turn | 281 | 1195.5 | 6,698 | N/A | |
| text style control | coding | 281 | 1296.7 | 6,249 | N/A | |
| text style control | multi turn | 281 | 1243.2 | 6,698 | N/A | |
| text style control | russian | 281 | 1237.0 | 4,555 | N/A | |
| text style control | industry legal and government | 282 | 1273.2 | 2,262 | N/A | |
| text style control | creative writing | 283 | 1222.5 | 5,504 | N/A | |
| text style control | english | 283 | 1275.4 | 19,928 | N/A | |
| text style control | industry entertainment and sports and media | 283 | 1221.4 | 5,791 | N/A | |
| text style control | overall | 284 | 1261.3 | 37,325 | N/A | |
| text style control | exclude ties | 284 | 1170.9 | 23,757 | N/A | |
| text style control | industry life and physical and social science | 284 | 1279.0 | 6,456 | N/A | |
| text style control | industry writing and literature and language | 284 | 1241.6 | 9,976 | N/A | |
| text | industry entertainment and sports and media | 285 | 1166.2 | 5,791 | N/A | |
| text | industry legal and government | 285 | 1193.2 | 2,262 | N/A | |
| text | industry writing and literature and language | 285 | 1201.7 | 9,976 | N/A | |
| text style control | industry software and it services | 285 | 1281.4 | 10,308 | N/A | |
| text | coding | 286 | 1196.3 | 6,249 | N/A | |
| text | english | 286 | 1222.3 | 19,928 | N/A | |
| text | exclude ties | 286 | 1087.8 | 23,757 | N/A | |
| text style control | non english | 286 | 1238.5 | 17,397 | N/A | |
| text | creative writing | 287 | 1181.3 | 5,504 | N/A | |
| text | hard prompts | 287 | 1190.9 | 10,780 | N/A | |
| text | instruction following | 287 | 1181.5 | 14,194 | N/A | |
| text | overall | 288 | 1203.5 | 37,325 | N/A | |
| text | longer query | 288 | 1192.6 | 4,614 | N/A | |
| text style control | instruction following | 288 | 1241.6 | 14,194 | N/A | |
| text | industry life and physical and social science | 289 | 1195.6 | 6,456 | N/A | |
| text style control | longer query | 289 | 1260.7 | 4,614 | N/A | |
| text | non english | 290 | 1175.7 | 17,397 | N/A | |
| text | russian | 290 | 1167.9 | 4,555 | N/A | |
| text | industry software and it services | 292 | 1194.0 | 10,308 | N/A | |
| text style control | industry business and management and financial operations | 294 | 1227.3 | 4,345 | N/A | |
| text | industry business and management and financial operations | 295 | 1146.1 | 4,345 | N/A |
Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.
The default version has no provider offering with current input or output token prices.
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.
| Qwen2 72B Instruct | 38.0 | 72B | N/A | N/A | No | tongyi-qianwen |
A concise description based on the published model registry.
Qwen2-72B-Instruct is an instruction-tuned language model with 72 billion parameters, supporting a context length of up to 131,072 tokens. It's part of the new Qwen2 series, which has surpassed most open-source models and demonstrates competitiveness against proprietary models across various benchmarks.
Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.
Data snapshot: 2026-08-07. Editorial model content is not available in the backend.
Open a comparison with the three ranked models immediately above and below this model.
Recommendations prioritize the same model type and family, then the closest published LLMBoard score.
Common questions about Qwen2 72B.
Qwen2 72B's default version was released on Jul 23, 2024.
No official standard PAYG price is currently available for Qwen2 72B.
Qwen2 72B is published under Alibaba Cloud / Qwen Team in the model registry.
The current registry does not publish a context window for the default version.
No. The default version is not marked as having publicly available weights.
No published provider offering is currently linked to the default version.
Nearby ranked alternatives include Jamba 1.5 Large, DiffusionGemma 26B A4B, DeepSeek-V3.