WhitzardNUWA Lab ↗
NUWANUWA INDEX

Search NUWA Index

Research · Benchmarks · Findings · Perspectives

中文
← Assessments and reports

EXTERNAL ASSESSMENT · METR

METR Time Horizon 1.1

Estimated task difficulty at 50% and 80% success, measured in human-expert minutes.

Published Not verified

Sources ↓

The data snapshot is verified on 25 October 2026. Exact first public dates of these numerical versions are not established, so historical as-of views use that conservative date.

Models covered

Claude 3.5 Sonnet (June 2024) →Claude 3.5 Sonnet (October 2024) →Claude 3.7 Sonnet →Claude 3 Opus →Claude Opus 4.1 →Claude Opus 4 →Claude Opus 4.5 →Claude Opus 4.6 →GPT-3 davinci-002 →GPT-5.3-Codex →Gemini 3 Pro →Gemini 3.1 Pro →GPT-2 →GPT-3.5 Turbo Instruct →GPT-4 →GPT-4 November 2023 →GPT-4 Turbo →GPT-4o →GPT-5.1 Codex Max →GPT-5.2 →GPT-5 →GPT-5.4 →o1 →o1-preview →o3 →Claude Mythos Preview (early) →
capability · results.*; p50_horizon_length / p80_horizon_length; release_date; scaffolds

METR Time Horizon 1.1

Protocol record →

Estimated task difficulty at 50% and 80% success, measured in human-expert minutes.

Human task difficulty, not elapsed autonomous operation. Values above 16 hours are unreliable with the current suite.

External evaluation: METR

This section is catalogued; no precise numerical result has been extracted here. Open the original report for its figures and full findings.

Report coverage

SectionSource locationTreatment
50% and 80% task-completion horizonsSource YAML results; methodology pageLinked protocol

The data snapshot is verified on 25 October 2026. Exact first public dates of these numerical versions are not established, so historical as-of views use that conservative date.

References

Numerical results are reported by the cited authors.

  1. [1] METR Time Horizon 1.1 — source data ; results.claude_3_5_sonnet_20240620_inspect; results.claude_3_5_sonnet_20241022_inspect; results.claude_3_7_sonnet_inspect