Quick Overview
- GPT-5.4, OpenAI’s most advanced professional model, became available on March 5, 2026
- Financial professionals gain access to integrated tools connecting FactSet and Third Bridge platforms
- Direct integration now available within Microsoft Excel and Google Sheets environments
- The release positions OpenAI as a direct competitor to Anthropic’s Claude for Financial Services
- ChatGPT Plus, Team, and Pro members can begin using the model immediately
On Thursday, March 5, 2026, OpenAI introduced GPT-5.4, its newest artificial intelligence model designed specifically with professional applications in mind. The update features specialized capabilities tailored for finance industry users.
The updated system can create spreadsheets, documents, and slide presentations with significantly reduced iterations. It performs web searches to compile data and deliver responses to sophisticated queries.
This release incorporates specialized financial capabilities that establish connections with FactSet Research Systems and Third Bridge data services. These features target professionals conducting financial research and developing investment documentation.
The model now functions directly within Microsoft Excel and Google Sheets applications. OpenAI simultaneously released a ChatGPT add-in specifically designed for Excel.
This launch intensifies competition with Anthropic, which previously introduced Claude for Financial Services. Both organizations are vying for enterprise clients prepared to invest in premium AI solutions.
Performance Metrics and Testing Results
During internal evaluations focusing on investment banking spreadsheet operations, GPT-5.4 achieved an 87.3% success rate, representing a substantial improvement from GPT-5.2’s 68.4% score. When human evaluators compared presentations, they selected GPT-5.4 output 68% of the time over its predecessor.
Using GDPval, a benchmark that evaluates AI performance across 44 different professional roles, GPT-5.4 equaled or exceeded human professional standards in 83% of test cases. The previous GPT-5.2 version achieved 70.9% on identical assessments.
The model recorded 75% accuracy on OSWorld-Verified, a benchmark measuring desktop navigation capabilities using visual inputs and pointer commands. Human participants scored 72.4% on the same evaluation.
OpenAI characterizes this as their most accurate model to date. Incorrect statements occur 33% less frequently than with GPT-5.2.
Access and Cost Structure
GPT-5.4 is becoming accessible to ChatGPT Plus, Team, and Pro members starting today under the designation GPT-5.4 Thinking. The preceding GPT-5.2 Thinking version will remain operational for three additional months before discontinuation on June 5, 2026.
Enterprise and Edu subscription holders can activate early access via administrative controls. GPT-5.4 Pro becomes available for Pro and Enterprise tier subscribers.
API usage costs begin at $2.50 per million input tokens and $15 per million output tokens. These rates exceed GPT-5.2 pricing, which stood at $1.75 and $14 respectively.
The system accommodates up to one million tokens of contextual data in Codex. Regular API calls handle 272,000 tokens, with expanded requests incurring double the standard rate.
According to OpenAI, GPT-5.4 delivers superior token efficiency compared to GPT-5.2, potentially offsetting higher per-token costs through reduced overall usage.
The model can be accessed through Codex, the company’s AI development tool, and via the standard API using the identifier gpt-5.4.
Get 3 Free Stock Ebooks
Discover top-performing stocks in AI, Crypto, and Technology with expert analysis.
- Top 10 AI Stocks - Leading AI companies
- Top 10 Crypto Stocks - Blockchain leaders
- Top 10 Tech Stocks - Tech giants
