Computer task execution on OSWorld (verified)
64.8Office Task ScoreCoAct-1
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| CoAct-1Steps=100 steps2025.08 | 64.8 | 61.6 | 71.82 | 75 | 47.87 | 59.93 | |
| CoAct-1Steps=150 steps2025.08 | 64.8 | 66.51 | 71.82 | 75 | 47.87 | 60.76 | |
| CoAct-1Steps=50 steps2025.08 | 62.91 | 59.43 | 69.89 | 70.83 | 42.37 | 56.38 | |
| GTA-1-7B w/ o3Steps=100 steps2025.08 | 55.68 | 64.74 | 61.2 | 62.5 | 38.34 | 53.07 | |
| Agent S2.5 w/ o3Steps=100 steps2025.08 | 54.23 | 55.8 | 75.42 | 75 | 44.06 | 55.98 | |
| Agent S2.5 w/ o3Steps=50 steps2025.08 | 52.81 | 55.8 | 75.42 | 75 | 39.53 | 54.21 | |
| Jedi-7B w/ o3Steps=50 steps2025.08 | 50.1 | 65.25 | 68.65 | 54.17 | 34.97 | 50.65 | |
| GTA-1-7B w/ o3Steps=50 steps2025.08 | 48.58 | 52.31 | 77.84 | 58.33 | 37.05 | 48.59 | |
| Jedi-7B w/ o3Steps=100 steps2025.08 | 47.89 | 64.37 | 75.92 | 50 | 35.27 | 50.98 | |
| CoAct-1Steps=15 steps2025.08 | 47.18 | 42.3 | 47.74 | 66.67 | 23.82 | 39.81 | |
| Jedi-7B w/ o3Steps=15 steps2025.08 | 45.84 | 57.49 | 60.95 | 50 | 20.43 | 42.37 | |
| Agent S2.5 w/ o3Steps=15 steps2025.08 | 42.85 | 44.61 | 57.1 | 70.83 | 17.82 | 38.98 | |
| OpenCUASteps=50 steps2025.08 | 30.06 | 42.31 | 58.28 | 47.83 | 16.79 | 33.76 | |
| OpenCUASteps=100 steps2025.08 | 30.06 | 38.89 | 60.7 | 52.17 | 18.1 | 33.84 | |
| UI-TARS-1.5 (7B)Steps=15 steps2025.08 | 27.19 | 27.99 | 61.45 | 34.78 | 5.38 | 25.76 | |
| OpenCUASteps=15 steps2025.08 | 26.69 | 33.25 | 51.57 | 43.48 | 10.41 | 28.12 | |
| UI-TARS-1.5 (7B)Steps=50 steps2025.08 | 26.51 | 31.41 | 48.91 | 25 | 9.77 | 25.08 | |
| OpenAI CUA 4oSteps=100 steps2025.08 | 25.04 | 39.19 | 55.43 | 58.33 | 18.48 | 31.38 | |
| UI-TARS-1.5 (7B)Steps=100 steps2025.08 | 25.01 | 31.07 | 46.99 | 29.17 | 8.8 | 25.41 | |
| OpenAI CUA 4oSteps=50 steps2025.08 | 23.56 | 38.43 | 52.09 | 70.83 | 15.86 | 31.19 | |
| OpenAI CUA 4oSteps=15 steps2025.08 | 22.17 | 37.65 | 41.22 | 45.83 | 10.75 | 26.01 | |
| OpenAI o3Steps=100 steps2025.08 | 17.23 | 26.29 | 38.79 | 62.5 | 16.53 | 23 | |
| OpenAI o3Steps=50 steps2025.08 | 11.5 | 19.78 | 30.1 | 37.5 | 11.82 | 17.17 | |
| OpenAI o3Steps=15 steps2025.08 | 1.45 | 8.02 | 12.29 | 37.5 | 11.82 | 9.09 |