Neurometric Leaderboard

neurometric Harbor Frontier Bench

Integrated Multi-Agent Analysis

Codex Gemini CLI Claude Code

Suite Selection

10 Active

Polyglot coding benchmark

200 Tasks

Codex GPT-5.4

75.4% n=1

Graduate-level science Q&A (diamond subset)

198 Tasks

Codex GPT-5.4

91.9% n=1

American Invitational Mathematics Examination

60 Tasks

Codex GPT-5.4

96.7% n=1

General AI Assistants benchmark

165 Tasks

Codex GPT-5.4

79.3% n=1

Live competitive programming benchmark

100 Tasks

Codex GPT-5.4

91.8% n=1

BigCodeBench Hard Complete

Hard code generation with complete specs

145 Tasks

Codex GPT-5.4

47.0% n=1

SWE-Bench Multilingual

Multilingual SWE-Bench tasks

300 Tasks

Codex GPT-5.4

50.7% n=1

Gemini CLI Gemini 3.1 Pro Preview

54.4% n=1

Claude Code Claude Opus 4.6

54.1% n=1

Quantum Circuit Bench

Quantum circuit reasoning benchmark

28 Tasks

Codex GPT-5.4

47.6% n=1

Gemini CLI Gemini 3.1 Pro Preview

46.4% n=1

Claude Code Claude Opus 4.6

42.9% n=1

Legacy code modernization benchmark (e.g. COBOL → modern languages)

10 Tasks

Codex GPT-5.4

0.0% n=1

Gemini CLI Gemini 3.1 Pro Preview

50.0% n=1

Claude Code Claude Opus 4.6

30.0% n=1

ARC-AGI 2 reasoning tasks

167 Tasks

Codex GPT-5.4

19.8% n=1