What's new in GPT-5.4?
What this update adds and why it matters
OpenAI’s latest frontier model advances the company’s push into professional, knowledge‑work AI. The release bundles faster reasoning modes, expanded context and better integrations aimed at workflows that combine text, code, tools and documents.
Key technical changes
- A much larger context window — up to 1 million tokens — lets the model hold entire long reports, books or multi‑file projects in memory at once.
- Native computer use: the model can directly interact with files, run code or manipulate applications in supported environments, closing the gap between suggestion and action.
- Improved tool calling and workflow support tailored to agents and multi‑step tasks.
Product and tiering details
OpenAI launched GPT‑5.4 in multiple variants. The company positioned this family as its most capable for professional work and introduced both a "Thinking" edition for deeper reasoning and a higher‑priced "Pro" tier. Public price points published by coverage include a lower tier charged at a few dollars per million tokens for inputs and outputs, while the Pro SKU is priced substantially higher for enterprise and high‑throughput use.
What it changes for users and businesses
Organizations that rely on long documents, complex chains of tools, or automated agents should see immediate benefits: better memory of context, fewer repeated prompts, and smoother handoffs between a model and external systems. The native computer use feature means the model can execute or orchestrate steps—drafting spreadsheets, formatting presentations, or running scripts—reducing manual work.
Limitations and risks
- OpenAI claims reductions in factual errors compared with prior models, but no model is perfect; verification remains necessary.
- Native computer control raises new safety and governance questions about automated actions and access controls.
Why it matters
Taken together, the technical and commercial choices in this release accelerate the shift from chatbots to agentic AIs that can act across apps and documents. That raises productivity potential but also increases the urgency of controls, auditing and clear usage policies for enterprises rolling these capabilities into production.