OpenAI has officially expanded its model lineup with the release of GPT-5.4 and GPT-5.4-pro, marking a significant leap in context handling and specialized task performance. These models feature a massive one-million-token context window, allowing them to process and reference the equivalent of several thick novels in a single prompt. This capacity is particularly useful for developers and researchers who need the system to maintain consistency across massive codebases or extensive document sets without losing track of information.

The update shows a clear strategic shift toward high-stakes business and financial applications. In internal testing, the model achieved an 87.3% success rate on complex spreadsheet modeling tasks typical of junior investment banking analysts, a sharp rise from the 68.4% recorded by its predecessor. This improvement suggests that the model is becoming increasingly adept at structural reasoning and multi-step data manipulation, areas where previous versions often struggled with accuracy.

Interestingly, GPT-5.4 has overtaken the previous coding-specific model, GPT-5.3-Codex, across all primary performance metrics. This raises questions about whether the developer-centric 'Codex' branch has now been merged into the main model line. While the capabilities are impressive, the cost remains a significant factor for the Pro version; individual high-fidelity image generations can cost over a dollar and take several minutes of processing time to complete.

OpenAI has officially expanded its model lineup with the release of GPT-5.4 and GPT-5.4-pro, marking a significant leap in context handling and specialized task performance. These models feature a massive one-million-token context window, allowing them to process and reference the equivalent of several thick novels in a single prompt. This capacity is particularly useful for developers and researchers who need the system to maintain consistency across massive codebases or extensive document sets without losing track of information.

The update shows a clear strategic shift toward high-stakes business and financial applications. In internal testing, the model achieved an 87.3% success rate on complex spreadsheet modeling tasks typical of junior investment banking analysts, a sharp rise from the 68.4% recorded by its predecessor. This improvement suggests that the model is becoming increasingly adept at structural reasoning and multi-step data manipulation, areas where previous versions often struggled with accuracy.

Interestingly, GPT-5.4 has overtaken the previous coding-specific model, GPT-5.3-Codex, across all primary performance metrics. This raises questions about whether the developer-centric 'Codex' branch has now been merged into the main model line. While the capabilities are impressive, the cost remains a significant factor for the Pro version; individual high-fidelity image generations can cost over a dollar and take several minutes of processing time to complete.