Model

o3

OpenAI · closed weights

openai.com

Knowledge & Reasoning

Knowledge & Reasoning
BenchmarkScoreMetricHarnessSourceConfidenceLink
CL-bench17.8overallN/AIndependentlowepoch.ai
Contradiction Benchmark for Review Evaluation99.9accuracyN/AIndependentmediumarxiv.org
CritPt1.4accuracyN/AIndependentlowepoch.ai
EnigmaEval13.1accuracyN/AIndependentmediumlabs.scale.com
GDPval30.8win_rateN/AIndependentlowepoch.ai
GPQA Diamond81.8accuracyN/AIndependenthighepoch.ai
HLE20.3accuracyN/AIndependentlowepoch.ai
SimpleQA Verified49.4accuracyN/AIndependenthighepoch.ai

Math

Math
BenchmarkScoreMetricHarnessSourceConfidenceLink
FrontierMath25.2accuracyN/AIndependentmediumpaperswithcode.com

Coding

Coding
BenchmarkScoreMetricHarnessSourceConfidenceLink
Aider Polyglot81.3percent_correctN/AIndependentmediumaider.chat
GSO8.8score_opt@1OpenHandsIndependentmediumgso-bench.github.io
SWE-bench Verified62.3resolved_rateN/AIndependenthighepoch.ai

Agentic / Tool Use

Agentic / Tool Use
BenchmarkScoreMetricHarnessSourceConfidenceLink
DeepResearch Bench45.2average_scoreN/AIndependentmediumdrb.futuresearch.ai
OSWorld23.0scoreo3 (100 steps)Independentmediumos-world.github.io
OSWorld17.2scoreo3 (50 steps)Independentmediumos-world.github.io
OSWorld9.1scoreo3 (15 steps)Independentmediumos-world.github.io

Multimodal

Multimodal
BenchmarkScoreMetricHarnessSourceConfidenceLink
FG-Sudoku0.8accuracyN/AIndependentmediumarxiv.org
GeoBench74.0acw_country_%N/AIndependentmediumgeobench.org
R-GroundBench67.5accuracyN/AIndependentmediumarxiv.org

Abstract Reasoning

Abstract Reasoning
BenchmarkScoreMetricHarnessSourceConfidenceLink
ARC-AGI-26.5scoreN/AIndependentlowepoch.ai

Reproduced by the community

Loading community runs…