Term
GPT-5.4
GPT-5.4 is OpenAI's reasoning model from March 5, 2026 — the first mainline model with the coding capabilities of GPT-5.3-codex, native computer-use controls and 1M-token context.
GPT-5.4 — explained in more detail
GPT-5.4 launched on March 5, 2026, initially in the Thinking and Pro variants. On March 17 the smaller GPT-5.4 mini (available to free-tier users) and GPT-5.4 nano (API only) followed. It is the first mainline reasoning model that ships with the frontier coding capabilities of GPT-5.3-codex — and in Codex and the API the first general-purpose model with native, state-of-the-art computer-use controls.
What GPT-5.4 delivers
- 1M-token context — matching the long-context tier of GPT-4.1 and Gemini.
- Native computer use — direct control of screens, apps and tools, no detours.
- Factual accuracy: individual claims 33% less likely to be false, full responses 18% less likely to contain errors than GPT-5.2 (on de-identified user error flags).
- Four sizes — Thinking, Pro, mini, nano — covering premium reasoning tasks down to high-volume API pipelines.
Where it fits
GPT-5.4 positions itself as the reasoning workhorse between the fast GPT-5.3 Instant line and the flagship GPT-5.5, which followed only six weeks later. A specialised GPT-5.4-Cyber variant was released for security teams.
Discover more
GPT-6 Astra Found Questions My First Security Review Missed
I used GPT-6 Astra and Fable 5.1 as independent reviewers for RLS, API and tenant-isolation checks. The useful part was the structured cross-review.
GlossaryGPT Astra
GPT-6 Astra is OpenAIs flagship from September 2026 — the first model at the Critical Preparedness tier, focused on computer use and cybersecurity.
EncyclopediaComparing AI models — who builds what and how to choose
The major model families in 2026 at a glance. Who builds Claude, GPT, Gemini, Llama, Mistral, DeepSeek, Qwen — and which model to pick when.