AI computer-use benchmarks (OSWorld, Tbench) have broken through human level

AI computer-use benchmarks (OSWorld, Tbench) have broken through human level | you've got the OS world verified and the Tathon benchmarks because OpenAI just bought OpenClaw. And now those benchmarks are actually just broken through human level. So AIs can use the computers better than humans.

you've got the OS world verified and the Tathon benchmarks because OpenAI just bought OpenClaw. And now those benchmarks are actually just broken through human level. So AIs can use the computers better than humans.

Beneficiaries (23)

Adverse (6)

{
  "nia": false,
  "url": "https://www.youtube.com/watch?v=d__HRChE2ZE",
  "mode": "THESIS",
  "role": "Host",
  "context": "you've got the OS world verified and the Tathon benchmarks because OpenAI just bought OpenClaw. And now those benchmarks are actually just broken through human level. So AIs can use the computers better than humans.",
  "to_year": 2026,
  "verbatim": "you've got the OS world verified and the Tathon benchmarks because OpenAI just bought OpenClaw. And now those benchmarks are actually just broken through human level. So AIs can use the computers better than humans.",
  "conv_cues": "broken through; better than humans",
  "direction": "HAPPEN",
  "from_year": 2026,
  "timeframe": "Now",
  "conv_level": "HIGH",
  "milestones": [
    {
      "kind": "llm_pre_event",
      "label": "Claude Opus 4.6 narrowly exceeds OSWorld human baseline at 72.7%",
      "source": "https://nerdleveltech.com/gpt-5-4-beats-humans-computer-use-ai-agents — Claude Opus 4.6 72.7%",
      "status": "hit",
      "weight": 0.4,
      "ordinal": -8,
      "source_id": null,
      "confidence": 0.95,
      "source_url": "https://nerdleveltech.com/gpt-5-4-beats-humans-computer-use-ai-agents",
      "expected_date": "2026-02-28",
      "observed_date": "2026-02-28",
      "research_origin": "deep_research",
      "measurement_criterion": "Claude Opus 4.6 scores 72.7% on OSWorld vs 72.4% human-expert baseline, validating thesis crossover"
    },
    {
      "kind": "llm_pre_event",
      "label": "GPT-5.4 hits 75.0% on OSWorld — first clear superhuman margin",
      "source": "https://nerdleveltech.com/gpt-5-4-beats-humans-computer-use-ai-agents — GPT-5.4 75.0%",
      "status": "hit",
      "weight": 0.4,
      "ordinal": -7,
      "source_id": null,
      "confidence": 0.95,
      "source_url": "https://nerdleveltech.com/gpt-5-4-beats-humans-computer-use-ai-agents",
      "expected_date": "2026-03-05",
      "observed_date": "2026-03-05",
      "research_origin": "deep_research",
      "measurement_criterion": "GPT-5.4 scores >=75% on OSWorld, exceeding human baseline by clear margin (>=2pp)"
    },
    {
      "kind": "llm_pre_event",
      "label": "Stanford AI Index 2026 documents OSWorld 12% to 66.3% accuracy jump",
      "source": "https://hai.stanford.edu/ai-index/2026-ai-index-report/technical-performance — Stanford AI Index",
      "status": "overdue",
      "weight": 0.4,
      "ordinal": -6,
      "source_id": null,
      "confidence": 0.92,
      "source_url": "https://hai.stanford.edu/ai-index/2026-ai-index-report/technical-performance",
      "expected_date": "2026-04-15",
      "miss_emitted_at": "2026-05-02T22:07:21.384228+00:00",
      "miss_emitted_by": "metadata_milestone_sweep",
      "research_origin": "deep_research",
      "measurement_criterion": "Stanford AI Index 2026 published showing OSWorld accuracy gain from ~12% to >=66% in 12 months"
    },
    {
      "kind": "prereq",
      "label": "Nvidia quadrupled chip production output while only doubling human headcount — achieved by deploying AI coding tools (Cursor, Claude Code) a",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -5,
      "source_id": "SEM_012",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "Training runs costing $10 billion for a single model will commence sometime in 2025.",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -4,
      "source_id": "SEM_008",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "2025 will be the definitive year that agentic systems finally hit the mainstream.",
      "status": "hit",
      "weight": 0.5,
      "ordinal": -3,
      "source_id": "SEM_042",
      "expected_date": "2026-04-29",
      "observed_date": "2026-04-29"
    },
    {
      "kind": "prereq",
      "label": "Recursive self-improvement is already happening now (no longer t
... (truncated)