Particle.news
Download on the App Store

OpenAI Launches GPT-5.4 With Native Computer Use and 1M-Token Context

The frontier release targets professional, agent‑driven tasks through a new Thinking mode that shows its plan for mid‑response steering.

Overview

  • GPT‑5.4 is rolling out across ChatGPT, the API, and Codex, with GPT‑5.4 Thinking for Plus, Team, and Pro users and a pricier Pro variant for Pro and Enterprise.
  • OpenAI’s first mainline model with native computer use can issue keyboard and mouse commands from screenshots and write code to operate apps.
  • The model supports up to a one‑million‑token context window and introduces tool search to cut token usage in large tool ecosystems.
  • OpenAI claims higher factuality versus GPT‑5.2, with 33% fewer false individual claims and 18% fewer errors, and top scores on OSWorld‑Verified, WebArena Verified, and APEX‑Agents.
  • Earlier this week, GPT‑5.3 Instant became ChatGPT’s default with reduced refusals and lower hallucination rates, while GPT‑5.2 remains in a legacy picker until June 3, 2026.