OpenAI launches GPT-5.4 with Pro and Thinking versions

OpenAI has launched GPT-5.4, described as its most capable and efficient frontier model for professional work, available in standard, GPT-5.4 Thinking (reasoning), and GPT-5.4 Pro (high performance) versions. The API version supports context windows up to 1 million tokens and boasts improved token efficiency over predecessors. Benchmark results show record scores in various tests, including OSWorld-Verified, WebArena Verified, and Mercor’s APEX-Agents, where it excels at long-horizon deliverables like financial models and legal analysis. Furthermore, GPT-5.4 demonstrates a 33% reduction in individual claim errors and an 18% overall reduction in response errors compared to GPT 5.2. The launch also introduced 'Tool Search' for API calls, which optimizes tool usage by allowing models to look up definitions as needed, and a new safety evaluation confirming that deception is less likely in the Thinking version.

(Source：TechCrunch)