OpenAI has officially released GPT-5.4, the latest iteration in its series of advanced language models, now available across multiple platforms including the API, Codex, and ChatGPT. The release includes two primary versions: GPT-5.4 Thinking and GPT-5.4 Pro, each optimized for specific performance and reasoning capabilities.
<h3>Key Features and Improvements</h3>
1. Enhanced Reasoning and Coding:
GPT-5.4 bridges advances in reasoning, coding, and agentic workflows, offering unprecedented performance in professional tasks. It combines coding capabilities inherited from GPT-5.3-Codex with improved support for complex workflows involving spreadsheets, documents, and presentations.
2. Large Context Window:
The model supports a context window of up to 1 million tokens, allowing for extensive content retention and application in large-scale projects.
3. Mid-Response Steering:
A notable feature of GPT-5.4 is its ability to let users influence the model’s output mid-response, enhancing interactivity and control, available on Android and the web, with iOS support to follow soon.
4. Improved Efficiency and Reduced Errors:
Compared to its predecessor GPT-5.2, the model achieves a 33% reduction in factual errors. It also introduces a new tool search system that significantly reduces token usage in multi-tool setups, boosting efficiency.
5. Integration and Availability:
<ul>
<li>Microsoft Foundry: GPT-5.4 is set for deployment in Microsoft Foundry, targeting enterprise applications with capabilities geared towards consistent reasoning and reduced latency.</li>
<li>Snowflake Cortex AI: Available in private preview, integrating with Cortex AI Functions and REST API, GPT-5.4 supports developer workflows and intelligence tasks within Snowflake’s secure environment.</li>
<li>GitHub Copilot: Released for use in various coding platforms, GPT-5.4 enhances logical reasoning in coding tasks.</li>
</ul>
6. Pricing and Accessibility:
Pricing for GPT-5.4 starts at $2.50 per million input tokens and $15.00 per million output tokens for the standard model, with the Pro version priced at $30.00 and $180.00, respectively. Despite being OpenAI’s most expensive model, its increased token efficiency is expected to offset costs.
<h3>Conclusion</h3>
OpenAI’s GPT-5.4 represents a significant advancement in AI-driven reasoning and professional task execution, characterized by improved error reduction, expanded context capabilities, and enhanced integrative functionalities for various enterprise and developer environments. The model is currently rolling out across ChatGPT and the Codex app, setting a new benchmark for AI capabilities.

Collections

OpenAI has released GPT-5.4 in two variants — GPT-5.4 Thinking and GPT-5.4 Pro — available via API, Codex, and ChatGPT. Key improvements include a 1 million token context window, 33% fewer factual errors compared to GPT-5.2, mid-response steering for interactive control, and a new tool search system that reduces token usage in multi-tool setups. The model integrates with Microsoft Foundry, Snowflake Cortex AI, and GitHub Copilot for enterprise and developer workflows. Pricing starts at $2.50 per million input tokens for the standard model, with the Pro version at $30.00 per million input tokens.

OpenAI Unveils GPT-5.4 with Enhanced Capabilities and Broader Integration