{ "question": "How often do AI agents succeed at real tasks?", "answer": "Top AI agents succeeded on 77.3% of real-world tasks in Terminal-Bench in early 2026, up from 20% in 2025, and on about 66% of real computer tasks in OSWorld, according to the Stanford AI Index 2026. That still means about one failed attempt in three on OSWorld, where people in the original 2024 study completed over 72.36% of tasks.", "period": "2026", "key_numbers": [ { "value": "77.3%", "label": "Agent success on real-world tasks in Terminal-Bench, up from 20% in 2025", "note": "Stanford AI Index 2026", "stat": "ai-agents/agent-success-terminal-bench" }, { "value": "About 66%", "label": "Agent success on real computer tasks in OSWorld, up from 12%", "note": "Stanford AI Index 2026", "stat": "ai-agents/agent-success-osworld" }, { "value": "72.36%", "label": "OSWorld tasks people completed in the benchmark's 2024 study", "note": "OSWorld paper, arXiv, April 2024" }, { "value": "93%", "label": "Cybersecurity problems solved by AI agents, up from 15% in 2024", "note": "Stanford AI Index 2026" }, { "value": "About 17 hours", "label": "Length of software task the most capable model tested finishes half the time", "note": "METR, model released Apr 2026; METR calls measurements above 16 hours unreliable" } ], "tables": [ { "id": "by-benchmark", "title": "AI agent success on benchmark tasks (Stanford AI Index 2026)", "columns": [ "Benchmark", "What it tests", "Earlier", "Latest (2026 AI Index)" ], "rows": [ [ "Terminal-Bench", "Real-world tasks in a computer terminal", "20% (2025)", "77.3%" ], [ "OSWorld", "Real computer tasks across operating systems", "12%", "About 66%" ], [ "Cybersecurity agent tasks", "Solving cybersecurity problems", "15% (2024)", "93%" ], [ "SWE-bench Verified", "Fixing real software issues (coding)", "60%", "Near 100% a year later" ], [ "Real household tasks (robots)", "Folding clothes, washing dishes and similar", "Not given", "12%" ] ], "row_ids": [ "terminal-bench", "osworld", "cybersecurity", "swe-bench-verified", "household-robots" ], "note": "Best scores reported in the AI Index 2026 (13 Apr 2026). Benchmark scores are lab tests of the best systems, not the success rate of agents deployed in businesses. The AI Index does not give the start year for the OSWorld 12%.", "source": "stanford-ai-index" }, { "id": "osworld-baseline", "title": "OSWorld when it was launched: people against the best model (April 2024)", "columns": [ "Who", "Share of OSWorld tasks completed" ], "rows": [ [ "People", "Over 72.36%" ], [ "Best AI model tested", "12.24%" ] ], "row_ids": [ "osworld-humans", "osworld-launch-model" ], "note": "OSWorld paper (arXiv 2404.07972, first submitted 11 Apr 2024). The models mainly struggled with recognising on-screen elements (GUI grounding) and operational knowledge. Two years later the AI Index puts the best agents at about 66%, close to the human level.", "source": "arxiv" }, { "id": "history", "title": "How long a software task the most capable AI agents can finish half the time (METR time horizon)", "columns": [ "Model release", "Model", "50% time horizon (human expert time)", "80% time horizon" ], "rows": [ [ "Mar 2023", "GPT-4", "About 4 minutes", "About 1 minute" ], [ "May 2024", "GPT-4o", "About 7 minutes", "About 1 minute" ], [ "Dec 2024", "o1", "About 39 minutes", "About 7 minutes" ], [ "Feb 2025", "Claude 3.7 Sonnet", "About 1 hour", "About 12 minutes" ], [ "Apr 2025", "o3", "About 2 hours", "About 30 minutes" ], [ "Aug 2025", "GPT-5", "About 3 hours 23 minutes", "About 38 minutes" ], [ "Nov 2025", "Claude Opus 4.5", "About 4 hours 53 minutes", "About 49 minutes" ], [ "Dec 2025", "GPT-5.2", "About 5 hours 52 minutes", "About 66 minutes" ], [ "Feb 2026", "Claude Opus 4.6", "About 12 hours", "About 70 minutes" ], [ "Apr 2026", "Claude Mythos Preview (early)", "About 17 hours (METR: measurements above 16 hours are unreliable)", "About 3 hours 6 minutes" ] ], "row_ids": [ "horizon-2023-03", "horizon-2024-05", "horizon-2024-12", "horizon-2025-02", "horizon-2025-04", "horizon-2025-08", "horizon-2025-11", "horizon-2025-12", "horizon-2026-02", "horizon-2026-04" ], "note": "METR Time Horizon 1.1 estimates (metr.org/time-horizons), a selection of the models METR marks as the state of the art when released, with model names as METR writes them; rounded from METR's point estimates in minutes. The 50% horizon is the length of task, measured by how long it takes a human expert, at which the agent is predicted to succeed half the time. Confidence intervals are wide (Claude Opus 4.6: about 5 to 61 hours), and METR says measurements above 16 hours are unreliable with its current task suite, which applies to the Claude Mythos Preview (early) figure. Software tasks only. METR stopped actively updating the page on 8 Sep 2026.", "source": "metr" } ], "sources": [ { "name": "AI Index Report", "publisher": "Stanford HAI", "type": "research", "url": "https://hai.stanford.edu/" }, { "name": "arXiv preprint server", "publisher": "Cornell University (arXiv)", "type": "study", "url": "https://arxiv.org/" }, { "name": "METR research notes", "publisher": "METR (Model Evaluation and Threat Research)", "type": "study", "url": "https://metr.org/" } ], "content_changed_on": "2026-10-31", "page_url": "https://digitaltreed.com/statistics/ai-agents/ai-agent-success-rate", "csv_url": "https://digitaltreed.com/statistics/ai-agents/ai-agent-success-rate.csv" }