Alibaba Cloud / Qwen Team model product
Qwen2.5-72B-Instruct is an instruction-tuned 72 billion parameter language model, part of the Qwen2.5 series. It is designed to follow instructions, generate long texts (over 8K tokens), understand structured data (e.g., tables), and generate structured outputs, especially JSON. The model supports multilingual capabilities across over 29 languages.
Updated Aug 10, 2026. Default version: Qwen2.5 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 Qwen2.5 72B Instruct.
| AlignBench | 0.8 | 1 | 4 | 100.0% | C | |
| MT-Bench | 0.9 | 1 | 12 | 100.0% | C | |
| MBPP | 0.9 | 5 | 37 | 88.9% | C | |
| Arena Hard | 0.8 | 6 | 26 | 80.0% | C | |
| MultiPL-E | 0.8 | 7 | 13 | 50.0% | C | |
| GSM8k | 1.0 | 10 | 48 | 80.8% | C | |
| MATH | 0.8 | 16 | 71 | 78.6% | C | |
| MMLU-Redux | 0.9 | 32 | 48 | 34.0% | C | |
| HumanEval | 0.9 | 33 | 76 | 57.3% | C | |
| LiveBench | 0.5 | 35 | 38 | 8.1% | C | |
| LiveCodeBench | 0.6 | 37 | 73 | 50.0% | C | |
| IFEval | 0.8 | 42 | 65 | 35.9% | C | |
| MMLU-Pro | 0.7 | 81 | 129 | 37.5% | C | |
| GPQA | 0.5 | 181 | 233 | 22.4% | C |
Preference and agent-evaluation signals from published Arena datasets.
| text | japanese | 190 | 1180.2 | 788 | N/A | |
| text | french | 197 | 1278.8 | 340 | N/A | |
| text style control | japanese | 200 | 1210.6 | 788 | N/A | |
| text | korean | 204 | 1188.1 | 527 | N/A | |
| text style control | french | 206 | 1319.7 | 340 | N/A | |
| text | spanish | 209 | 1254.3 | 347 | N/A | |
| text style control | korean | 209 | 1229.3 | 527 | N/A | |
| text | industry mathematical | 215 | 1290.5 | 4,565 | N/A | |
| text style control | spanish | 215 | 1291.2 | 347 | N/A | |
| text | german | 218 | 1233.4 | 926 | N/A | |
| text | math | 222 | 1282.6 | 5,415 | N/A | |
| text style control | german | 222 | 1265.6 | 926 | N/A | |
| text style control | industry mathematical | 222 | 1309.6 | 4,565 | N/A | |
| text | chinese | 224 | 1271.5 | 3,329 | N/A | |
| text style control | chinese | 226 | 1321.2 | 3,329 | N/A | |
| text | expert | 227 | 1244.5 | 2,397 | N/A | |
| text style control | industry legal and government | 227 | 1330.5 | 2,386 | N/A | |
| text style control | math | 227 | 1296.3 | 5,415 | N/A | |
| text | industry legal and government | 228 | 1288.0 | 2,386 | N/A | |
| text | coding | 231 | 1292.9 | 6,688 | N/A | |
| text | longer query | 233 | 1282.0 | 6,008 | N/A | |
| text | russian | 233 | 1263.1 | 5,776 | N/A | |
| text | industry software and it services | 234 | 1293.2 | 10,448 | N/A | |
| text style control | coding | 234 | 1355.5 | 6,688 | N/A | |
| text | hard prompts | 235 | 1271.0 | 10,545 | N/A | |
| text style control | russian | 237 | 1296.4 | 5,776 | N/A | |
| text | industry medicine and healthcare | 238 | 1257.7 | 1,956 | N/A | |
| text style control | expert | 238 | 1293.8 | 2,397 | N/A | |
| text | instruction following | 239 | 1254.4 | 16,363 | N/A | |
| text | industry business and management and financial operations | 240 | 1252.6 | 4,656 | N/A | |
| text style control | industry software and it services | 240 | 1343.4 | 10,448 | N/A | |
| text | multi turn | 241 | 1271.8 | 7,250 | N/A | |
| text style control | longer query | 241 | 1317.0 | 6,008 | N/A | |
| text style control | multi turn | 241 | 1298.8 | 7,250 | N/A | |
| text | hard prompts english | 242 | 1281.3 | 6,208 | N/A | |
| text | industry life and physical and social science | 244 | 1275.3 | 6,932 | N/A | |
| text | non english | 245 | 1251.6 | 18,845 | N/A | |
| text style control | hard prompts english | 245 | 1327.8 | 6,208 | N/A | |
| text style control | instruction following | 245 | 1292.2 | 16,363 | N/A | |
| text style control | hard prompts | 246 | 1317.5 | 10,545 | N/A | |
| text style control | industry business and management and financial operations | 246 | 1299.2 | 4,656 | N/A | |
| text style control | industry medicine and healthcare | 246 | 1312.8 | 1,956 | N/A | |
| text style control | non english | 247 | 1284.9 | 18,845 | N/A | |
| text | exclude ties | 250 | 1189.7 | 24,444 | N/A | |
| text | overall | 251 | 1269.2 | 39,406 | N/A | |
| text | english | 253 | 1282.8 | 20,561 | N/A | |
| text style control | industry life and physical and social science | 253 | 1318.3 | 6,932 | N/A | |
| text style control | industry writing and literature and language | 253 | 1282.0 | 10,827 | N/A | |
| text style control | exclude ties | 255 | 1237.7 | 24,444 | N/A | |
| text style control | overall | 257 | 1302.9 | 39,406 | N/A | |
| text | industry writing and literature and language | 258 | 1246.9 | 10,827 | N/A | |
| text style control | english | 260 | 1316.2 | 20,561 | N/A | |
| text style control | creative writing | 263 | 1254.3 | 5,684 | N/A | |
| text | creative writing | 264 | 1221.6 | 5,684 | N/A | |
| text | industry entertainment and sports and media | 266 | 1211.6 | 6,207 | N/A | |
| text style control | industry entertainment and sports and media | 274 | 1244.9 | 6,207 | 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.
| Abacus | Qwen/Qwen2.5-72B-Instruct | global | $0.11 | $0.38 | 128K | |
| Kilo Gateway | qwen/qwen-2.5-72b-instruct | global | $0.36 | $0.40 | 32.8K | |
| OpenRouter | qwen/qwen-2.5-72b-instruct | global | $0.36 | $0.40 | 32.8K | |
| NovitaAI | qwen/qwen-2.5-72b-instruct | global | $0.38 | $0.40 | 32K | |
| Alibaba (China) | qwen2-5-72b-instruct | global | $0.574 | $1.72 | 131.1K | |
| Alibaba | qwen2-5-72b-instruct | global | $1.4 | $5.6 | 131.1K |
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.
| Qwen2.5 72B Instruct | 53.1 | 72.7B | 131.1K | 8.2K | No | Qwen |
A concise description based on the published model registry.
Qwen2.5-72B-Instruct is an instruction-tuned 72 billion parameter language model, part of the Qwen2.5 series. It is designed to follow instructions, generate long texts (over 8K tokens), understand structured data (e.g., tables), and generate structured outputs, especially JSON. The model supports multilingual capabilities across over 29 languages.
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 Qwen2.5 72B.
Qwen2.5 72B's default version was released on Sep 19, 2024.
Qwen2.5 72B's official API price is $1.4 per million input tokens and $5.6 per million output tokens via Alibaba. The lowest tracked third-party offer starts at $0.11 input and $0.38 output via Abacus.
Qwen2.5 72B is published under Alibaba Cloud / Qwen Team in the model registry.
The default version has a 131.1K token context window.
No. The default version is not marked as having publicly available weights.
6 published provider offerings are linked to the default version.
Nearby ranked alternatives include Llama 3.3 70B, Qwen3 Next 80B A3B Thinking, Qwen3.5 35B A3B.