News
White Paper
Why Enterprise AI Fails at Scale
Key Takeaways
- GPT-5.4 introduces a one-million-token context window.
- Native computer-use capabilities allow the model to operate software and websites like a human user.
- OpenAI claims GPT-5.4 outperforms professionals in 83% of tasks across 44 occupations on the GDPval benchmark.
OpenAI released GPT-5.4 on March 5, 2026. According to company officials, the new model combines advances in reasoning, coding and agentic workflows.
GPT-5.4 includes coding capabilities from GPT-5.3-Codex while enhancing performance across tools, software environments and tasks involving spreadsheets, presentations and documents. OpenAI also released GPT-5.4 Pro for users requiring maximum performance on complex tasks.
The one-million-token context window puts GPT-5.4 on par with offerings from Google and Anthropic. According to OpenAI, the model's “extreme” thinking mode allows it to apply more compute to difficult questions, aimed at reducing errors on longer tasks.
Table of Contents
- GPT-5.4 Feature Breakdown
- AI Moves Closer to Autonomous Office Work
- OpenAI Locks In Chips, Cloud and Capital
- About OpenAI
GPT-5.4 Feature Breakdown
OpenAI claims GPT-5.4 delivers improvements across multiple benchmarks and capabilities:
| Feature | How It Works |
|---|---|
| One-million-token context | Supports long-horizon task planning and execution |
| Extreme thinking mode | Applies more compute to complex questions |
| Native computer use | Enables agents to operate software via mouse and keyboard |
| GPT-5.4 Thinking | Provides upfront plan, allows mid-response adjustments |
| Enhanced document editing | Improved spreadsheet, presentation and document creation |
On GDPval, which tests agent performance across 44 occupations, OpenAI reports GPT-5.4 matches or exceeds industry professionals in 83% of comparisons, compared to 70.9% for GPT-5.2.
On OSWorld-Verified, measuring desktop navigation, GPT-5.4 achieved 75% success, exceeding human performance of 72.4%.
AI Moves Closer to Autonomous Office Work
"GPT-5.4 is the best model we've ever tried. It's now top of the leaderboard on our APEX-Agents benchmark... It excels at creating long-horizon deliverables such as slide decks, financial models and legal analysis, delivering top performance while running faster and at a lower cost than competitive frontier models."
- Brendan Foody
CEO, Mercor
OpenAI has made major advancements in autonomous work in 2025 and 2026:
- February 2026: OpenAI dropped its Codex agent, which offers extended-context capabilities by processing more than 7 million tokens from a single initial prompt.
- December 2025: The company released GPT-5.2, which introduced advances in professional knowledge work automation.
- August 2025: OpenAI debuted ChatGPT Agent, which performs complex, multi-step tasks by interacting with websites, running code and navigating workflows.
Alongside these new developments, ChatGPT usage has risen dramatically. Data shows users sent 5x more messages in 2025 compared to 2024 — totaling 18 billion weekly.
OpenAI Locks In Chips, Cloud and Capital
On the enterprise expansion side, OpenAI has deployed billions in acquisitions, infrastructure partnerships and product launches over the last two years.
In September of last year, the company acquired experimentation platform Statsig for $1.1 billion. Three months later, it acquired Neptune for under $400 million to enhance model governance.
OpenAI also diversified its compute infrastructure through a massive partnership with NVIDIA, where the chipmaker agreed to invest up to $100 billion and become OpenAI's key AI chip supplier starting in 2026.
Additionally, in February of this year, Amazon announced a $50 billion investment and exclusive third-party cloud distribution rights for OpenAI's Frontier platform.
Webinar
Mar
25
Content Leaders Collective: Navigating Content Decisions at Scale
Discover how content leaders are modernizing content operations, avoiding costly missteps and preparing for scale and AI.
Register
Webinar
On demand
Content Strategy Leaders Live: Scaling for Speed, Complexity and AI in High Tech
A candid roundtable on how high-tech leaders are rethinking content at scale.
Watch Now
Webinar
On demand
Do More with Less: Modernizing the Cloud Contact Center for 2026
Learn how to leverage cloud platforms without adding a single hire to personalize every customer interaction.
Watch Now
Webinar
On demand
Cut the Noise: Deploying AI That Actually Moves the Needle
Learn how to turn AI experimentation into concrete revenue operations.
Watch Now
Webinar
On demand
Ditch the Desk Phones: How Modern Teams Drive AI-First Communications
Find out how one team finally pulled the plug on a legacy phone system. And built something smarter.
Watch Now
Webinar
On demand
Rebrand. Migrate. Optimize. How to Do It All (Without Slowing Down)
Cresta leveled up site speed, design flexibility and marketer sanity (in record time). Find out how.
Watch Now
About OpenAI
OpenAI, founded in 2015, develops AI technologies for enterprise and developer audiences. Key offerings include large language models such as GPT and Codex, as well as the DALL-E image generation system.