GPT-5.4 Is A Substantial Upgrade
The new model costs more but aims to reclaim the lead from Claude Opus 4.6 in key areas.
OpenAI has officially launched GPT-5.4, positioning it as a significant step up from its predecessor, GPT-5.2. According to CEO Sam Altman, the new model brings major improvements in knowledge work, web search, and introduces native computer use capabilities—meaning the AI can now interact with and control software. It supports a massive 1 million tokens of context, allows users to steer its responses mid-generation, and is rolling out across ChatGPT and the API. Early analysis suggests this release is OpenAI's move to reclaim its competitive edge, as many felt Anthropic's Claude Opus 4.6 had been dominating for general use.
Initial user reports and a gestalt of community reactions indicate GPT-5.4 excels at assembling facts, coding, and providing comprehensive rundowns. Its writing quality and perceived personality have notably improved, making it more useful for creative and editorial tasks. However, the upgrade comes with a cost increase; one analysis showed API expenses rising from approximately $2,304 to $2,951 for a comparable workload, though it remains cheaper than Opus in its maximum performance mode. The model is not a revolutionary leap in core general capability, but a focused enhancement on common use cases and internal metrics, with the 'model relations' community noting substantial progress in multi-agent interactions.
- Native computer use capability allows the model to interact with and control software directly.
- Supports 1 million tokens of context, a major increase for handling long documents and conversations.
- API costs rose ~28% in one analysis, but the model aims to compete directly with Claude Opus 4.6.
Why It Matters
Professionals gain a more capable AI for coding and knowledge tasks, but must evaluate cost versus performance against leading competitors.