GPT-5.4 (Low): benchmark results

GPT-5.4 evaluated at the low reasoning-effort setting. Provider: OpenAI. Released 2026-03-05. Access: API.

Aggregate Rating 1740 ± 14, rank #261 of 2168 rated models, from 78 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
SentinelBench (sleep Tool)68Task success rate (%) over 100 time-evolving monitoring task100
SentinelBench (wait_for Tool)75Task success rate (%) over 100 time-evolving monitoring task100
AA-Omniscience Index - Software Engineering (SWE) - Julia76Omniscience Index96
K-Bench - Risk (Therapeutic v0)94.38Risk Score (%, Therapeutic v0 prompt)95.2
AA-Omniscience Index - Software Engineering (SWE) - R50Omniscience Index93.7
K-Bench (Therapeutic v0)98.51Overall (%, Therapeutic v0 prompt)93.5
AA-Omniscience Index - Software Engineering (SWE) - TypeScript76.67Omniscience Index93.4
AA-Omniscience Index - Software Engineering (SWE) - Python73.5Omniscience Index93.3
AA-Omniscience Index - Software Engineering (SWE) - JavaScript72.73Omniscience Index93.2
AA-Omniscience Index - Software Engineering (SWE)61.2Omniscience Index91.2
AA-Omniscience Index - Software Engineering (SWE) - Go64Omniscience Index91.1
AA-Omniscience Index - Software Engineering (SWE) - Kotlin52Omniscience Index90.8

Interactive version: theaggregate.ai/model?slug=gpt-5-4-low · How It Works · Data refreshed daily, snapshot 2026-10-20.