OpenAI officially launched GPT-5.4 on March 5, 2026, positioning it as the most capable and efficient frontier model for professional work. Available across ChatGPT, the API, and Codex, the model comes in three configurations: a standard version, GPT-5.4 Thinking for complex reasoning tasks, and GPT-5.4 Pro for users who require maximum performance. The new model integrates the advanced coding capabilities of GPT-5.3-Codex with improved reasoning, native computer-use functionality, and enhanced performance across professional knowledge work tasks such as spreadsheets, presentations, and documents.
One of GPT-5.4's most significant claims is its reduction in hallucinations and factual errors. According to OpenAI, individual claims made by the model are 33% less likely to be false compared to GPT-5.2, while full responses are 18% less likely to contain any errors. On the GDPval benchmark — which evaluates AI performance across 44 occupations spanning the top nine industries contributing to U.S. GDP — GPT-5.4 matches or outperforms industry professionals in 83% of comparisons, a notable jump from GPT-5.2's 70.9% win rate. The model also scored 87.3% on an internal investment banking spreadsheet benchmark, compared to 68.4% for GPT-5.2.
In the realm of computer use and software development, GPT-5.4 sets new benchmarks. On OSWorld-Verified, which measures a model's ability to complete tasks in a desktop environment using screenshots and keyboard/mouse inputs, GPT-5.4 achieved a 75% success rate — surpassing human performance at 72.4% and dramatically outpacing GPT-5.2's 47.3%. On SWE-Bench Pro, a rigorous software engineering evaluation, GPT-5.4 scored 57.7%, edging past GPT-5.3-Codex's 56.8%. The model also introduces tool search in the API, which reduced total token usage by 47% in benchmark testing while maintaining the same accuracy — a major efficiency gain for large-scale agentic workflows.
Pricing for GPT-5.4 is set at $2.50 per million input tokens and $15 per million output tokens, higher than GPT-5.2's $1.75 and $14 respectively. OpenAI argues the cost increase is offset by the model's improved token efficiency, which requires fewer tokens to accomplish equivalent tasks. In ChatGPT, GPT-5.4 Thinking replaces GPT-5.2 Thinking for Plus, Team, and Pro subscribers starting today, while GPT-5.2 Thinking will remain accessible under Legacy Models until June 5, 2026. Enterprise and Edu plan users can enable early access through admin settings.