Alibaba (Qwen)
Qwen 2.5 72B
openrouter/qwen-2.5-72bContext
33K
Input / 1M tokens
$0.3600
Output / 1M tokens
$0.4000
Cached input / 1M tokens
— (bills at input price)
Cache write / 1M tokens
— (bills at input price)
About this model
Qwen 2.5 72B
Qwen 2.5 72B is Alibaba's flagship dense open-weights model. Strong general-purpose chat with particular strength on math, code, and multilingual tasks. The 131K context window comfortably handles long documents.
Key facts
- Maker
- Alibaba (Qwen)
- Context window
- 33K tokens
- Max output
- 16K tokens
- Lane
- Guarded: the provider has confirmed it does not train on your data
- Modality
- Text
- Input price
- $0.3600 per 1M tokens
- Output price
- $0.4000 per 1M tokens
- Processing region(s)
- us
- Route
- Via OpenRouter
- Cheapest plan
- Available from the Pro plan
- Price comparison
- Priced below 206 of 230 pro-band guarded models on Anoman (blended price: 1 part input to 3 parts output).
Released
2024-09
Training cutoff
2024-06
Parameters
72.7B (dense)
Ways to call
Ways to call Qwen 2.5 72B
Each model ID below is callable through the Anoman AI API. Price and processing region depend on the route.
| Model ID | Route | Input / output per 1M tokens | Region(s) | Lane | Context |
|---|---|---|---|---|---|
| openrouter/qwen-2.5-72b | Via OpenRouter | $0.3600 / $0.4000 | us | Guarded | 33K |
| qwen-2.5-72b | Via SiliconFlow's API | $0.3600 / $0.4000 | cn | Guarded | 33K |
| qwen-2.5-72b-openrouter-backup | Via OpenRouter | $0.3600 / $0.4000 | us | Guarded | 33K |
Best for
Use cases
- General-purpose production chat
- Multilingual workloads (29 languages supported)
- Code generation + technical documentation
- Long-document RAG over PDFs / contracts / codebases
Strengths
What it does well
- 131K context window — among the largest open-weights options
- Strong on Asian-language benchmarks (CMMLU, JMMLU, KMMLU)
- Native structured output + tool calling
- Available open-weights for fully air-gapped deploys
- Quality-per-dollar in our catalog's top tier
Limitations
Know the trade-offs
- Knowledge cutoff 2024 — fresh facts require RAG or search
- Vision variants exist (Qwen 2.5 VL) but this row is text-only
- Trails frontier closed-source on the hardest reasoning tasks
Benchmarks
Published scores
Scores from official model cards and public leaderboards. Higher is better unless noted.
| Benchmark | Score | Measures |
|---|---|---|
| MMLU | 86.1 | General knowledge across 57 subjects |
| HumanEval | 86.6 | Python code generation, pass@1 |
| MATH | 83.1 | Mathematics, mixed difficulty |
| GPQA-Diamond | 49.0 | Graduate-level science questions |
Code example
Chat completion with Qwen 2.5 72B
from openai import OpenAI
client = OpenAI(
base_url="https://api.anoman.io/v1",
api_key="anm-sk-..."
)
response = client.chat.completions.create(
model="openrouter/qwen-2.5-72b",
messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)Alternatives
Similar models in our catalog
Same maker first, then the closest price to Qwen 2.5 72B in the Anoman catalog.