Qwen2.5-VL 72B Instruct
| Released | 2025-01-26 |
|---|---|
| Price per 1M | $1.01 in · $1.01 out |
| Overall score | 35.8 · rank 213 |
| Context window | 131,072 tokens |
| Provider | Alibaba |
| Weights | Open weights |
Scores
| Dimension | Score | Rank | Benchmarks |
|---|---|---|---|
| Overall: Balanced | 35.8 | 213 | 10 |
| Overall: Coding | 35.2 | 208 | 5 |
| Overall: Reasoning | 37.1 | 207 | 4 |
| Overall: Agentic work | 34.8 | 217 | 5 |
| Overall: Frontend | 37.6 | 187 | 2 |
| Agents | 34.2 | 175 | 3 |
| Reasoning | 37.2 | 214 | 2 |
| Math | 37.0 | 173 | 2 |
| General | 34.7 | 176 | 1 |
| Multimodal | 37.6 | 113 | 2 |
| Value vs price | −12.0 | 183 |
Evidence
Agents 34.2 ±4.0
| Benchmark | Result | Source | Date |
|---|---|---|---|
| METR task suite | 35.8% | Epoch AI | 2026-11-11 |
| BALROG | 16.2% | Epoch AI | 2026-11-11 |
| OSWorld-Verified | 5.0% | OSWorld-Verified | 2025-07-28 |
| OSWorld (v1) | 5.0% | Epoch AI | 2026-11-11 |
Reasoning 37.2 ±4.0
| Benchmark | Result | Source | Date |
|---|---|---|---|
| GPQA Diamond | 49.2% | Epoch AI | 2025-01-27 |
| Kagi Reasoning Benchmark | 36.0% | Kagi | 2026-11-11 |
Math 37.0 ±2.0
| Benchmark | Result | Source | Date |
|---|---|---|---|
| MATH Level 5 | 63.2% | Epoch AI | 2025-01-27 |
| OTIS Mock AIME 2024-2025 | 8.1% | Epoch AI | 2025-02-25 |
General 34.7 ±4.1
| Benchmark | Result | Source | Date |
|---|---|---|---|
| LMArena Text | 1,303 | LMArena | 2026-11-09 |
Multimodal 37.6 ±4.0
| Benchmark | Result | Source | Date |
|---|---|---|---|
| LMArena Vision | 1,121 | LMArena | 2026-11-09 |
| GeoBench | 62.0% | Epoch AI | 2026-11-11 |
Results not used in the score
| Benchmark | Result | Source | Why not used |
|---|---|---|---|
| OSWorld (v1) | 4.4% | Epoch AI | A stronger configuration result was used |
| OSWorld-Verified | 4.4% | OSWorld-Verified | A stronger configuration result was used |
Names matched to this model
qwen2-5-72b, qwen2-5-72b-instruct, qwen2-5-vl-72b, qwen2-5-vl-72b-instruct