Qwen2.5-72B
Qwen2.5-72B from Alibaba (Qwen).
Shogo Agent Index
Estimated#91 of 9212.1range 8.5–15.7 · 14% of weight measured
Agentic coding
11
estimated
Long-horizon autonomy
20
50% measured
Knowledge-work agents
12
estimated
Computer use
2
40% measured
Security & context learning
9
estimated
Agentic benchmarks
| Benchmark | Score | Normalised | Index weight | Harness / source |
|---|---|---|---|---|
| SWE-bench Verified | est. | 8% | predicted 38 | |
| Terminal-Bench 2.0 | est. | 8% | predicted 2 | |
| DeepSWE | est. | 6% | predicted 4 | |
| FrontierSWE | est. | 5% | predicted 0 | |
| CursorBench | est. | 4% | predicted 12 | |
| GSO-Bench | est. | 4% | predicted 1 | |
| METR Time Horizons | 35.8% | 10% | source | |
| Vending-Bench 2 | est. | 6% | predicted 0 | |
| PostTrainBench | est. | 4% | predicted 8 | |
| APEX-Agents | est. | 8% | predicted 5 | |
| GDPval | est. | 7% | predicted 10 | |
| Remote Labor Index | est. | 5% | predicted 0 | |
| DeepResearch Bench | est. | 5% | predicted 38 | |
| OSWorld 2.0 | est. | 6% | predicted 0 | |
| OSWorld | 5.0% | 4% | qwen2.5-vl-72b-instruct (100 steps) | |
| Cybench | est. | 5% | predicted 9 | |
| CL-bench | est. | 5% | predicted 10 |
Other benchmarks
| Benchmark | Category | Score | Source |
|---|---|---|---|
| BALROG | Long-horizon | 16.2% | source |
| DTBench | Knowledge | 62.9% | source |
| GeoBench | Multimodal | 62.0% | source |
| GPQA Diamond | Reasoning | 49.1% | source |
| LMCA | Knowledge | 13.4% | source |
| MATH Level 5 | Math | 63.2% | source |
| MMLU | Legacy | 85.3% | source |
| OTIS Mock AIME | Math | 8.1% | source |
| TheAgentCompany | Knowledge work | 5.7% | source |