OpenAI launches GPT-5.4 with 1 million-token context and agentic workflows
OpenAI launches GPT-5.4 with 1 million-token context and agentic workflows
- OpenAI launched GPT-5.4 on March 5, 2026, positioning it as the most capable and efficient frontier model for professional applications by merging the Codex and GPT lineages. openai.com+1
- It offers a 1 million‑plus token context window (up to 1.05 million tokens) and multimodal text‑and‑image processing, enabling high‑context reasoning for complex problems. openai.com+1
- The model provides state‑of‑the‑art coding abilities, superior world knowledge, and improved token efficiency for tool calling and front‑end development. gaga.art+1
- A breakthrough native computer use feature lets GPT-5.4 interact directly with software interfaces, supporting sophisticated agentic workflows as a digital assistant. gaga.art+1
- In real‑world tests it matches or surpasses professionals in 83 % of knowledge‑work tasks and reduces factual errors by 33 %, while using fewer tokens than its predecessor. gaga.art+1
- Rolling out across ChatGPT, API, and Codex, it comes in GPT-5.4 Thinking for deep reasoning and GPT-5.4 Pro for enterprise, priced at $2.50 per million input tokens and $20 per million output tokens. openai.com+1
