GPT-6 Astra: New Features, Differences, Pricing & What’s Changed

OpenAI’s GPT-6 Astra is designed to take AI beyond answering questions and toward actually completing complex digital work.

OpenAI has introduced GPT-6 Astra, its newest flagship AI model, positioning it as a major advance in reasoning, computer interaction, coding, research and professional workflows.

The most interesting part of Astra is not simply that it is a newer GPT model. The bigger change is its focus on using computers and tools to accomplish tasks. Instead of only telling users what to do, the model is designed to navigate software, websites and other digital environments more like an active assistant.

OpenAI says Astra is now available across ChatGPT Work, Codex and the API, with capabilities spanning computer use, browsing, software engineering, cybersecurity, science and professional work.

GPT-6 Astra vs Previous GPT Models: What Has Changed?

The evolution from earlier GPT models to Astra can be understood as a shift from conversation to execution.

OpenAI describes Astra as state-of-the-art in computer use, browsing, professional work, software engineering, cybersecurity and science.

The difference is therefore less about a single flashy feature and more about how the model is intended to be used.

New Features in GPT-6 Astra

1. Computer Use

Computer use is arguably Astra's most important new capability.

The model is designed to interact with computers more directly, including navigating software and websites. Reports describing the launch say Astra can work across browsers, spreadsheets, websites and desktop applications.

This opens the door to tasks such as:

• Navigating websites
• Working with online applications
• Handling multi-step digital workflows
• Interacting with software
• Assisting with spreadsheets and documents
• Performing tasks that previously required constant human interaction

This is an important step toward AI agents that perform tasks rather than simply generate responses.

2. Stronger Coding

Astra is also aimed at advanced software engineering.

Instead of being limited to producing isolated snippets of code, the model is designed for more complex development workflows involving codebases, debugging and software-engineering tasks.

OpenAI lists software engineering among the areas in which Astra achieves state-of-the-art performance.

For developers, this could mean AI becoming more useful as a long-running engineering partner rather than simply a coding chatbot.

3. Better Research and Scientific Reasoning

Scientific research is another major target.

OpenAI says Astra has capabilities across science and complex professional work. Recent reporting has also highlighted the model's ability to contribute to difficult mathematical and scientific problems.

The significance is that AI is increasingly being positioned not only as a writing or productivity tool but as a system that can help researchers explore difficult technical questions.

4. Cybersecurity

Astra also represents a significant step in cybersecurity capabilities.

OpenAI's deployment-safety material says Astra is the first OpenAI model to meet its Critical cybersecurity capability threshold.

That makes cybersecurity both an opportunity and a safety concern. More capable AI can help defenders identify vulnerabilities and improve security, but the same capabilities can potentially be misused.

5. Better Professional Work

Astra is particularly aimed at professional users.

OpenAI says the model is built for demanding professional work, with improvements in reasoning, computer use, writing and design judgment.

This could make it useful for areas such as:

• Business analysis
• Software development
• Research
• Data-heavy workflows
• Technical writing
• Design-related tasks
• Financial and professional services

OpenAI has already incorporated GPT-6 Astra into a specialized ChatGPT offering for financial services, where it is being used for research, financial reasoning and content generation.

GPT-6 Astra Pricing

For developers using the API, currently reported pricing is approximately:

$10 per 1 million input tokens

$50 per 1 million output tokens

Recent independent and industry sources report the same $10/$50 per-million-token structure.

That makes Astra a premium frontier model rather than a low-cost model intended for every simple AI request.

The practical cost will depend heavily on how much input and output an application generates and how much reasoning or computer interaction a task requires.

For ordinary users, access can also depend on the ChatGPT plan or product through which Astra is offered. OpenAI has made Astra available through ChatGPT Work, Codex and the API, while reports indicate availability for higher-tier ChatGPT users as well.

Important: API token pricing and consumer subscription pricing are different things, so users should not assume that the $10/$50 API rates represent the monthly cost of using Astra inside ChatGPT.

What Is the Core Point of GPT-6 Astra?

The central idea behind Astra can be summed up in one sentence:

AI should not just tell you how to complete a task—it should increasingly be capable of completing the task itself.

This is the most important difference.

Traditional chatbots operate mainly through a question-and-answer model:

User → Question → AI → Answer

Astra is moving toward:

User → Goal → AI → Plan → Tools/Computer → Completed Task

That is a fundamental change in the role of an AI model.

Imagine asking:

"Research these five companies, compare their products, put the information into a spreadsheet and prepare a summary."

A conventional AI might provide the research and explain how to create the spreadsheet.

An agentic model aims to perform much more of that workflow itself.

That is why computer use, tool interaction and long-running professional tasks are so important to Astra.

Why GPT-6 Astra Matters

GPT-6 Astra arrives at a point where the AI industry is moving rapidly from generative AI toward agentic AI.

The first major wave of AI focused on generating:

• Text
• Images
• Code
• Summaries
• Answers

The next wave is increasingly focused on taking actions.

Astra is part of that transition.

For businesses, this could mean AI systems that perform repetitive digital work. For developers, it could mean more autonomous coding agents. For researchers, it could mean AI systems capable of exploring complex problems. And for ordinary users, it could eventually mean a digital assistant capable of handling entire workflows.

But greater autonomy also creates greater risks.

A model that can interact with computers is potentially much more powerful than one that can only generate text. OpenAI's emphasis on alignment and deployment safety reflects this growing concern.

Final Verdict

GPT-6 Astra is not simply another upgrade in the GPT series. Its bigger ambition is to make AI more capable of acting.

Its major developments include stronger reasoning, computer use, coding, browsing, scientific work, cybersecurity and professional workflows.

The model's reported $10 per million input tokens and $50 per million output tokens API pricing places it firmly in the premium category, making it most attractive for demanding applications where capability matters more than minimizing every token cost.

The real test, however, will not be benchmark scores alone.

The important question is whether Astra can reliably complete complicated real-world tasks while remaining safe, predictable and under human control.

If it can, GPT-6 Astra could represent an important step toward a future where AI is less like a chatbot—and more like a capable digital coworker.