AlibabaQwen2.5 72B
Best configuration ranks #469 of 610 on the BenchLeader Index at 42.1 ±4.6. Last measured 2 Sept 2026. Released 1 Sept 2024.
- Blended price
- $0.861/M
- $0.574 in · $1.72 out
- Output speed
- 10 tok/s
- First answer
- 7.92 s
- Context
- 131k
- Overall index42
- Reasoning41
- Coding37
- Agents & tools38
- Maths40
- Knowledge39
- Instruction following40
- Human preference47
- Composite39
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | default | Source |
|---|---|---|
| GPQA Diamond | 49.1%#202 | Epoch AI Benchmarking Hub |
| LMArena Hard Prompts | 1317#225 | LMArena |
| GPQA Diamond (AA)not in index | 49.1%#410 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 3.6%#502 | Artificial Analysis |
Coding
| Benchmark | default | Source |
|---|---|---|
| WeirdML | 16.0%#139 | WeirdML |
| LMArena Coding | 1355#216 | LMArena |
Agents & tools
| Benchmark | default | Source |
|---|---|---|
| Terminal-Bench Hard | 4.5%#270 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 34.5%#231 | Artificial Analysis |
Maths
| Benchmark | default | Source |
|---|---|---|
| OTIS Mock AIME | 8.1%#213 | Epoch AI Benchmarking Hub |
| MATH Level 5 | 63.2%#44 | Epoch AI Benchmarking Hub |
Knowledge
| Benchmark | default | Source |
|---|---|---|
| AA-Omniscience | -53.1#342 | Artificial Analysis |
Instruction following
| Benchmark | default | Source |
|---|---|---|
| IFBench | 36.9%#281 | Artificial Analysis |
Human preference
| Benchmark | default | Source |
|---|---|---|
| LMArena Text | 1303#232 | LMArena |
Composite
| Benchmark | default | Source |
|---|---|---|
| Epoch Capabilities Indexnot in index | 129#136 | Epoch AI Benchmarking Hub |
| AA Intelligence Index | 7.7#383 | Artificial Analysis |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| DeepInfra | 12 tok/s | 2.44 s | $0.360 | $0.400 | 33k | fp8 |
| NovitaAI | 7 tok/s | 13 s | $0.380 | $0.400 | 32k | bf16 |
Price history
Listed price per 1M tokens over time, as recorded by OpenRouter for the provider with the longest history.
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0007 | – | 39.5 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0079 | – | 1.2 min |
| Code edit | 6,000 / 1,500 | $0.0060 | – | 2.8 min |
| Agentic coding session | 60,000 / 4,000 | $0.041 | – | 7.2 min |
| Structured extraction | 2,000 / 200 | $0.0015 | – | 29.0 s |
See also
Data as of 9 Sept 2026. Compare with another model.