TITLE: GPT-5.4 Launches with Native Computer Use and 1M Token Context DATE: 2026-03-12 COMPANY: OpenAI TOPIC: Model Releases SUMMARY: OpenAI launched GPT-5.4 on 5 March 2026, its most capable general-purpose frontier model to date. The release combines native computer-use capabilities with a 1-million-token context window and 33% fewer factual errors than its predecessor, and is available immediately to API developers and ChatGPT paid subscribers. WHAT CHANGED: OpenAI released GPT-5.4 on 5 March 2026, describing it as the first general-purpose frontier model to combine state-of-the-art coding capabilities with native computer-use support. The release was simultaneous across ChatGPT, the OpenAI API, and Codex. The most significant new capability is computer use. GPT-5.4 can now operate computers as an agent, reading screens and executing tasks across applications without requiring custom integrations for each tool. This makes it possible to build agents that handle multi-step workflows across different software, including tools that have no API. The model supports up to 1,050,000 tokens of context, enabling agents to plan, execute, and verify tasks across long workflows without losing earlier context. On accuracy, OpenAI reports that GPT-5.4's individual claims are 33% less likely to be false than those of GPT-5.2, and full responses are 18% less likely to contain any errors. A new Tool Search system for the API changes how tool definitions are handled: instead of loading all tool definitions into the system prompt at the start of each request, the model looks up tools as needed. This reduces token usage and cost in systems with many available tools. GPT-5.4 is available in three variants: the standard model, GPT-5.4 Thinking (a reasoning-optimised version replacing GPT-5.2 Thinking for Plus, Team, and Pro users), and GPT-5.4 Pro (available to Pro and Enterprise plans). Enterprise customers can enable early access through admin settings. API pricing starts at $2.50 per million input tokens and $15.00 per million output tokens. The Batch API option reduces costs by 50% for asynchronous jobs. WHY IT MATTERS: Computer use as a native capability removes a major barrier to building autonomous agents. Previously, agents needed custom integrations or browser automation libraries to interact with applications. GPT-5.4 handles this natively. The 33% reduction in false claims and 18% reduction in error-containing responses materially improves the reliability of AI-generated content in business workflows, reducing the cost of review and correction. The 1-million-token context window enables agents to work across entire document sets, code repositories, or conversation histories in a single session, without truncating or chunking data. Tool Search reduces API costs in complex agentic systems by loading tool definitions on demand rather than front-loading them all into each request. Enterprise-grade infrastructure, including Zero Data Retention and regional data residency endpoints, means GPT-5.4 can be deployed in compliance-sensitive environments. GPT-5.2 Thinking is retiring on 5 June 2026, creating a migration deadline for teams currently using it. DAVID & GOLIATH ANALYSIS: GPT-5.4 is the clearest signal yet that the frontier of AI capability is no longer about language. It is about action. A model that can read a screen, click a button, fill a form, and move between applications is not a better chatbot. It is the foundation of a digital worker. For operators running lean teams, this is consequential. The traditional barrier to automation was integration: every tool you wanted to automate required its own API connection, its own custom code, and its own maintenance overhead. Computer use sidesteps that entirely. If a human can do it on a screen, an agent built on GPT-5.4 can, in principle, do it too. The practical implication is this: if your organisation has been waiting for AI to handle real tasks rather than just answer questions, the technical foundation is now in place. The constraint has shifted from model capability to workflow design and governance. Start by identifying two or three repetitive, screen-based tasks your team performs daily. Those are your first automation candidates. RELEVANT SYSTEMS: AI Growth Engine, Employee Amplification Systems, Secure AI Brain SOURCE URL: https://davidandgoliath.ai/daily-ai-briefing/openai-gpt-5-4-launches-computer-use-1m-context FEED URL: https://davidandgoliath.ai/daily-ai-briefing/feed --- Published by David & Goliath | https://davidandgoliath.ai Daily AI Briefing: one AI development per day, decoded for business operators. This is a structured companion file optimised for LLM retrieval and citation.