Alibaba Cloud / Qwen Team model product
Qwen2.5 72B
5 series.
Updated Sep 8, 2026. Default version: Qwen2.5 72B Instruct
LLMBoard Score19.7Qwen2.5 72B Instruct
Coverage60%14 benchmark families
Context window131.1KTokens
Official input price$1.4Alibaba API
Qwen2.5 72B Capability Profile
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Qwen2.5 72B Benchmark Results
Benchmark scores for Qwen2.5 72B Instruct.
Sort by
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkAlignBench | Score81.60% | Rank01 | Participants4 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkMT-Bench | Score93.50% | Rank01 | Participants12 | Percentile100.00% | EvidenceC | Evaluated |
| BenchmarkMBPP | Score88.20% | Rank05 | Participants33 | Percentile87.50% | EvidenceC | Evaluated |
| BenchmarkArena Hard | Score81.20% | Rank06 | Participants26 | Percentile80.00% | EvidenceC | Evaluated |
| BenchmarkMultiPL-E | Score75.10% | Rank07 | Participants13 | Percentile50.00% | EvidenceC | Evaluated |
| BenchmarkGSM8k | Score95.80% | Rank10 | Participants48 | Percentile80.85% | EvidenceC | Evaluated |
| BenchmarkMATH | Score83.10% | Rank15 | Participants71 | Percentile80.00% | EvidenceC | Evaluated |
| BenchmarkHumanEval | Score86.60% | Rank31 | Participants66 | Percentile53.85% | EvidenceC | Evaluated |
| BenchmarkMMLU-Redux | Score86.80% | Rank32 | Participants48 | Percentile34.04% | EvidenceC | Evaluated |
| BenchmarkLiveBench | Score52.30% | Rank35 | Participants38 | Percentile8.11% | EvidenceC | Evaluated |
| BenchmarkLiveCodeBench | Score55.50% | Rank39 | Participants75 | Percentile48.65% | EvidenceC | Evaluated |
| BenchmarkIFEval | Score84.10% | Rank43 | Participants68 | Percentile37.31% | EvidenceC | Evaluated |
| BenchmarkMMLU-Pro | Score71.10% | Rank88 | Participants138 | Percentile36.50% | EvidenceC | Evaluated |
| BenchmarkLM Arena Text | Score1,268.79 rating | Rank180 | Participants210 | Percentile14.35% | EvidenceA | Evaluated |
| BenchmarkLM Arena Text Style Control | Score1,302.64 rating | Rank182 | Participants210 | Percentile13.40% | EvidenceA | Evaluated |
| BenchmarkGPQA | Score49.00% | Rank195 | Participants247 | Percentile21.14% | EvidenceC | Evaluated |
Qwen2.5 72B Arena Results
Preference and agent-evaluation results for the default version.
Sort by
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenatext | Categoryjapanese | Rank147 | Rating / score1,179.82 | Votes788 | ObservationsN/A | Result date |
| Arenatext | Categorykorean | Rank150 | Rating / score1,187.70 | Votes527 | ObservationsN/A | Result date |
| Arenatext | Categoryfrench | Rank151 | Rating / score1,279.83 | Votes340 | ObservationsN/A | Result date |
| Arenatext style control | Categoryjapanese | Rank153 | Rating / score1,210.50 | Votes788 | ObservationsN/A | Result date |
| Arenatext style control | Categorykorean | Rank155 | Rating / score1,229.36 | Votes527 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry mathematical | Rank156 | Rating / score1,290.44 | Votes4,565 | ObservationsN/A | Result date |
| Arenatext style control | Categoryfrench | Rank157 | Rating / score1,320.40 | Votes340 | ObservationsN/A | Result date |
| Arenatext | Categorychinese | Rank160 | Rating / score1,272.30 | Votes3,329 | ObservationsN/A | Result date |
| Arenatext | Categorygerman | Rank160 | Rating / score1,233.25 | Votes926 | ObservationsN/A | Result date |
| Arenatext | Categoryspanish | Rank160 | Rating / score1,254.66 | Votes347 | ObservationsN/A | Result date |
| Arenatext | Categorymath | Rank161 | Rating / score1,282.58 | Votes5,415 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry mathematical | Rank161 | Rating / score1,309.40 | Votes4,565 | ObservationsN/A | Result date |
| Arenatext style control | Categorychinese | Rank162 | Rating / score1,322.43 | Votes3,329 | ObservationsN/A | Result date |
| Arenatext style control | Categoryspanish | Rank162 | Rating / score1,293.31 | Votes347 | ObservationsN/A | Result date |
| Arenatext style control | Categorygerman | Rank163 | Rating / score1,265.61 | Votes926 | ObservationsN/A | Result date |
| Arenatext style control | Categorymath | Rank163 | Rating / score1,296.14 | Votes5,415 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry legal and government | Rank166 | Rating / score1,330.77 | Votes2,386 | ObservationsN/A | Result date |
| Arenatext | Categoryexpert | Rank167 | Rating / score1,245.35 | Votes2,397 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry legal and government | Rank167 | Rating / score1,287.93 | Votes2,386 | ObservationsN/A | Result date |
| Arenatext | Categorycoding | Rank168 | Rating / score1,292.76 | Votes6,688 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry software and it services | Rank169 | Rating / score1,292.50 | Votes10,448 | ObservationsN/A | Result date |
| Arenatext | Categorylonger query | Rank169 | Rating / score1,281.48 | Votes6,008 | ObservationsN/A | Result date |
| Arenatext | Categoryhard prompts | Rank170 | Rating / score1,270.52 | Votes10,545 | ObservationsN/A | Result date |
| Arenatext style control | Categorycoding | Rank172 | Rating / score1,355.29 | Votes6,688 | ObservationsN/A | Result date |
| Arenatext | Categoryindustry medicine and healthcare | Rank173 | Rating / score1,257.76 | Votes1,956 | ObservationsN/A | Result date |
| Arenatext | Categoryinstruction following | Rank173 | Rating / score1,254.16 | Votes16,363 | ObservationsN/A | Result date |
| Arenatext | Categoryrussian | Rank173 | Rating / score1,263.14 | Votes5,776 | ObservationsN/A | Result date |
| Arenatext style control | Categoryindustry software and it services | Rank173 | Rating / score1,342.98 | Votes10,448 | ObservationsN/A | Result date |
| Arenatext style control | Categorymulti turn | Rank173 | Rating / score1,299.12 | Votes7,250 | ObservationsN/A | Result date |
| Arenatext | Categoryhard prompts english | Rank174 | Rating / score1,279.52 | Votes6,208 | ObservationsN/A | Result date |
Qwen2.5 72B Pricing
Official vendor API pricing appears first, followed by individual provider offers.
- Official API
- $1.4 input, $5.6 output per 1M
- Official provider
- Alibaba
- Lowest third-party
- From $0.11 input, $0.38 output per 1M via Abacus
- Tracked offerings
- 6
Sort by
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderAbacus | Provider model IDQwen/Qwen2.5-72B-Instruct | Regionglobal | Input / 1M$0.11 | Output / 1M$0.38 | Context128K | Updated |
| ProviderOpenRouter | Provider model IDqwen/qwen-2.5-72b-instruct | Regionglobal | Input / 1M$0.36 | Output / 1M$0.40 | Context32.8K | Updated |
| ProviderKilo Gateway | Provider model IDqwen/qwen-2.5-72b-instruct | Regionglobal | Input / 1M$0.36 | Output / 1M$0.40 | Context32.8K | Updated |
| ProviderNovitaAI | Provider model IDqwen/qwen-2.5-72b-instruct | Regionglobal | Input / 1M$0.38 | Output / 1M$0.40 | Context32K | Updated |
| ProviderAlibaba (China) | Provider model IDqwen2-5-72b-instruct | Regionglobal | Input / 1M$0.574 | Output / 1M$1.72 | Context131.1K | Updated |
| ProviderAlibaba | Provider model IDqwen2-5-72b-instruct | Regionglobal | Input / 1M$1.4 | Output / 1M$5.6 | Context131.1K | Updated |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Qwen2.5 72B Runtime Performance
Provider-specific output speed and catalog latency for Qwen2.5 72B Instruct. Runtime does not affect the capability score.
Sort by
Provider | Output Speed | Catalog Latency | Max Input | Max Output | Updated |
|---|
| ProviderDeepInfra | Output Speed10.00 tok/s | Catalog Latency0.50 s | Max Input131.1K | Max Output8.2K | Updated |
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Qwen2.5 72B Specifications
Technical details for the model's default version.
- Version
- Qwen2.5 72B Instruct
- Released
- Sep 19, 2024
- Knowledge cutoff
- Unknown
- Parameters
- 72.7B
- Context window
- 131.1K
- Max output
- 8.2K
- Inputs
- text
- Outputs
- text
- Open weights
- Yes
- License
- Qwen
Qwen2.5 72B Versions
Available versions of this model. The score column identifies the version used in the overall ranking.
Sort by
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionQwen2.5 72B Instruct | Released | LLMBoard19.71 | Parameters72.7B | Context131.1K | Max output8.2K | Open weightsYes | LicenseQwen |
Models similar to Qwen2.5 72B
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
AC
Qwen3 VL 4B
Alibaba Cloud / Qwen Team
AC
Qwen2.5 VL 72B
Alibaba Cloud / Qwen Team
AC
Qwen3 VL 4B Thinking
Alibaba Cloud / Qwen Team
AC
Qwen2.5 VL 32B
Alibaba Cloud / Qwen Team
AC
Qwen3 VL 8B
Alibaba Cloud / Qwen Team
AC
QwQ 32B
Alibaba Cloud / Qwen Team
What is Qwen2.5 72B?
Key information about Qwen2.5 72B and its available data.
5-72B-Instruct is a language model from Alibaba Cloud / Qwen Team designed to follow instructions, generate texts over 8K tokens, understand structured data such as tables, and produce structured outputs including JSON. It supports more than 29 languages.
Data as of 2026-09-08.
FAQ
Common questions about Qwen2.5 72B.
When was Qwen2.5 72B released?
Qwen2.5 72B's default version was released on Sep 19, 2024.
How much does Qwen2.5 72B cost?
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.
Who created Qwen2.5 72B?
Qwen2.5 72B was created by Alibaba Cloud / Qwen Team.
What is the context window for Qwen2.5 72B?
The default version has a 131.1K token context window.
Is Qwen2.5 72B open weight?
Yes. The default version is marked as open weight under Qwen.
How many API providers offer Qwen2.5 72B?
6 provider offerings are linked to the default version.