AI benchmarks and leaderboards are different things
Learn the difference between a benchmark, an evaluation run and a leaderboard, and use each to answer the right question.
LLM Stats
OSWorld 2.0
OSWorld 2.0 is a benchmark of 108 long-horizon, real-world computer-use workflows spanning everyday and professional tasks. Each task is an end-to-end workflow that takes human users a median of about 1.6 hours, scored with a binary-completion metric, and targets challenges such as dynamic environments, cross-source reasoning, and implicit-state inference.
Evaluation context and evidence
- Source
- LLM Stats
- Publisher
- Not recorded
- Benchmark release
- Not recorded
- Collected
- 2026-08-17
- Modality
- multimodal
These are source-reported observations, not independent reruns by this site. Check the task version, evaluation protocol, unit and reporting date before comparing models. Results from different setups are not a universal ranking.
Original sources
Reported results 5
Shown in source observation order, not ranked. Dates retain the precision provided by the source.
| Model | Value | Unit | Protocol | Reported date and basis | Source |
|---|---|---|---|---|---|
| Claude Opus 5 | 0.706 | Not recorded | Not recorded | 2026-07-24 · model_announcement | LLM Stats |
| GPT-5.6 Sol | 0.626 | Not recorded | Not recorded | 2026-07-09 · model_announcement | LLM Stats |
| GPT-5.6 Terra | 0.502 | Not recorded | Not recorded | 2026-07-09 · model_announcement | LLM Stats |
| Gemini 3.7 Flash | 0.479 | Not recorded | Not recorded | 2026-08-13 · model_announcement | LLM Stats |
| GPT-5.6 Luna | 0.456 | Not recorded | Not recorded | 2026-07-09 · model_announcement | LLM Stats |