Artificial intelligence · March 2026

GPT-5.4

A model that uses a desktop better than the average person.

GPT-5.4 shipped with built-in computer use and scored 75% on OSWorld-Verified, above the 72.4% human baseline. OpenAI reported 33% fewer factual errors than GPT-5.2.

See it in the living tree Wikipedia Source

Arcs above join it to what it stood on, arcs below to what it led to. The bar places this stretch on the tree's whole timeline.

What it built on 2

GPT-5.2 2025
Computer use agents 2024

What it led to 1

GPT-5.5 2026

Related measurements

The dashed line marks when it first worked.

Best AI score on GPQA Diamond Latest: 95.8 % correct (2026) Source: Epoch AI, Capabilities & benchmarking (benchmark_data.zip, gpqa_diamond.csv); Anthropic: Claude 3 public launch (March 4, 2024)

Training compute of notable AI models Source: Epoch AI, Notable AI Models (CC BY 4.0)