GPQA Diamond ↗94.6%
GPT-5.4 Pro (xhigh) · GPQA diamond · Best score (across scorers) · % ↗
OpenAI / MODEL EVIDENCE
GPT-5.4 Pro
14 benchmarks · 16 observations · 15 sources · released 2026-03-05
Scores at a glance
14 benchmarks · large dot: GPT-5.4 Pro; small dots: closest models where reported
GPT-5.4 Pro (xhigh) · FrontierMath-2025-02-28-Private · Best score (across scorers) · % ↗
GPT-5.4 Pro (xhigh) · FrontierMath-Tiers-1-3-v2-Private · Best score (across scorers) · % ↗
GPT-5.4 Pro (xhigh) · FrontierMath-Tier-4-v2-Private · Best score (across scorers) · % ↗
SimpleQA Verified ↗46.3%
GPT-5.4 Pro (xhigh) · SimpleQA Verified · Best score (across scorers) · % ↗
GPT-5.4 Pro (xhigh) · Chess Puzzles · Best score (across scorers) · % ↗
ARC-AGI-1 ↗94.5%
GPT-5.4 Pro (xhigh) · ARC-AGI · Score · % ↗
1 more reported result
94.5% · GPT-5.4 Pro (xhigh) · ARC-AGI · Score · % · Epoch AI · ARC-AGI ↗
SimpleBench ↗74.1%
GPT-5.4 Pro · SimpleBench · Score (AVG@5) · % ↗
WeirdML (v2) ↗57.44%
GPT-5.4 Pro (no thinking) · WeirdML · Accuracy · % ↗
ARC-AGI-2 ↗83.33%
GPT-5.4 Pro (xhigh) · ARC-AGI-2 · Score · % ↗
Humanity’s Last Exam ↗44.32%
GPT-5.4 Pro · HLE · Accuracy · % ↗
CritPt ↗30%
GPT-5.4 Pro (xhigh) · CritPt · Accuracy · % ↗
EnigmaEval ↗23.82%
GPT-5.4 Pro · EnigmaEval · Accuracy · % ↗
Epoch Capabilities Index (ECI) ↗159.57 ECI
GPT-5.4 Pro · General ECI · 26 October 2026 fit ↗
Closest models by score
Similar positions on shared benchmark leaderboards. Smaller gaps are closer.
Claude Sonnet 5.5 →Anthropic
5 of 6 shared benchmark comparisons
GPT-5.5 Pro →OpenAI
5 of 6 shared benchmark comparisons
Qwen3.8 Max →Qwen
5 of 6 shared benchmark comparisons
How similarity is measured
Mean absolute gap in percentile positions on shared benchmark settings; lower is closer. At least 60% overlap is required, with a minimum of three when available. Independent evaluations are used when there are at least three; otherwise published report comparisons are used. Repeated model observations within one setting are averaged before ranking. Writing-style traits and composite indices are excluded. This describes similarity, not an overall quality ranking.
Sources
Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.
- ARC Prize - Leaderboard ↗
- ARC-AGI-1 ↗
- ARC-AGI-2 ↗
- Chess Puzzles ↗
- CritPt v1 ↗
- FrontierMath Tier 4 (v2) ↗
- FrontierMath Tiers 1-3 (v1, old) ↗
- FrontierMath Tiers 1-3 (v2) ↗
- GPQA Diamond ↗
- Humanity's Last Exam ↗
- SimpleQA Verified ↗
- SimpleBench ↗
- AI benchmarks & capabilities ↗
- WeirdML ↗
- EnigmaEval ↗
Names used in source tables
GPT-5.4 Pro · GPT-5.4 Pro (no thinking) · GPT-5.4 Pro (xhigh)