GPT-5.4 (High) — benchmark results

GPT-5.4 evaluated at the high reasoning-effort setting. Provider: OpenAI. Released 2026-03-06. Access: API.

Unified ELO 1915 ± 23, rank #54 of 1776 rated models, from 82 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
APEX v167.2Score (%)100
CubeBench66.7Success Rate (%)100
GENSTRAT85Alpha (chips/game) (self-reported)100
LLM2014 Logic 2026-0378.85Median Score100
OpenCompass LLM - Language80.2Score (%)100
OpenCompass Language - Creation82.5Score (%)100
OpenCompass Language - NLP74.6Score (%)100
OpenCompass Reasoning - Academic52Score (%)100
Persuasion (Lechmazur)1.71Average Persuasion Strength100
VTB29.17Score (self-reported)100
UGI - Natural Intelligence71.25NatInt Score98.6
UGI - Writing68.82Writing Score98.4

Interactive version: theaggregate.ai/model?slug=gpt-5-4-high · How the rankings work · Data refreshed daily, snapshot 2026-07-22.