OpenAI has unveiled its latest advancements in artificial intelligence with the release of GPT-5.4 Thinking and Pro, enhancing the capabilities of its most powerful model. This announcement follows the earlier launch of GPT-5.3 Instant, which aimed to increase the naturalness of interactions with its popular chatbot, ChatGPT. The rollout of GPT-5.4 is set to gradually occur today across various platforms, marking a significant upgrade from the previous version, GPT-5.2, which was introduced to counter competitive pressures from Google’s Gemini in December.
The new model brings a suite of improvements focused on various operational areas. OpenAI has emphasized six key enhancements, including advancements in coding, document understanding, and instruction following, which are crucial for users in technical fields. The model also shows notable improvements in image perception and multimodal tasks, enabling it to manage more complex interactions between text and images. Additionally, it excels at long-running task execution and multi-step agent workflows, which are critical for applications requiring sustained processing.
Among the technical upgrades, the support for a 1 million token context window stands out, allowing users to analyze extensive codebases or large collections of documents in a single request. This feature significantly enhances the usability of the model for users dealing with substantial datasets or intricate projects. Furthermore, GPT-5.4 introduces capabilities for “built-in computer use,” enabling the model to interact directly with software to perform, verify, and rectify tasks in what OpenAI describes as a build-run-verify-fix loop.
OpenAI has also highlighted that GPT-5.4 is designed to support compaction, which extends the length of agent trajectories while preserving important context. This enhancement is particularly relevant for applications in customer service, analytics, and finance, where document-heavy and spreadsheet-heavy workflows are common. The model’s ability to synthesize information from multiple sources, especially for hard-to-locate data, enhances its utility for users requiring comprehensive answers and insights.
As part of this transition, GPT-5.4 Thinking will be available to subscribers of the Plus, Team, and Pro plans, effectively replacing the older GPT-5.2 Thinking, which will be phased out in three months. OpenAI’s gradual rollout strategy allows users to adapt to the new capabilities while ensuring a smooth transition.
With these enhancements, OpenAI continues to reinforce its position in the competitive landscape of AI development. The introduction of GPT-5.4 underscores the company’s commitment to pushing the boundaries of what is possible in AI, while addressing the needs of a diverse user base seeking sophisticated tools to enhance productivity and decision-making. As AI continues to evolve, the implications of these advancements will likely resonate across various sectors, reshaping interactions with technology and influencing the future of work.
See also
OpenAI Launches GPT-5.3 Instant, Reducing Cringe and Enhancing Accuracy
Synthetic Media Revolution: AI Deepfakes Raise Trust Issues, Demand New Literacy Skills
BABA-W Launches Qwen-Image 2.0 with Advanced Image Generation and Editing Features
Raspberry Pi 5 Launches Offline AI Assistant with Vision, Voice, and Image Generation Features
