OpenAI Rolls Out GPT-6 Astra, Its ‘Most Capable Model’ Yet

OpenAI has launched GPT-6 Astra, its latest frontier model, in a bid to push into AI systems that can use computers and execute tasks more autonomously.

OpenAI says that GPT-6 Astra is its most capable model yet. The model supports reasoning efforts ranging from low to maximum and comes with a 1.05 Mn-token context window and a maximum output of 1.28 Lakh tokens. 

As per the ChatGPT parent, Astra can autonomously navigate websites, interact with software, work with documents and spreadsheets, write and test code, and handle other multi-step workflows. The model is pitched as a significant upgrade for AI interface.

“Astra saturates FrontierMath Tier 4 with a 98% score, having already helped solve long-standing open problems⁠ in mathematics. Astra also saturates ARC-AGI-3 with a 99.9% score and ExploitBench with a 100% score. It also sets a new frontier on computer and browser use, handling the most demanding professional work with unmatched speed, accuracy, and judgment,” the company said. 

Initially, GPT‑6 Astra will be rolled out to a limited set of organisations and will become available to all ChatGPT Plus, Pro, Business, and Enterprise users, soon. 

GPT-6 Astra has an April 30, 2026 knowledge cutoff and supports tools including web search, file search, computer use, code execution and image generation. 

It is priced at $10 per Mn input tokens and $50 per Mn output tokens, with discounted cached-input pricing of $1 per Mn tokens.

The company also doubles down on agentic AI with its new model, a segment where it is competing with companies including Anthropic and Google. Instead of requiring users to break down a task into a series of prompts, these systems are designed to perform several steps on their own.

OpenAI claimed state-of-the-art results for Astra on benchmarks including Agents’ Last Exam, AutomationBench and ScreenSpot Pro, which measure an AI model’s ability to handle computer-based workflows.

“We believe it is the best model in the world for computer use, professional work, science, coding, cybersecurity, and more,” OpenAI CEO Sam Altman said in a social media post on X.

GPT-6 Astra Steps Up Coding, Science Capabilities

ChatGPT parent has also positioned GPT-6 Astra as its strongest model yet for software engineering.

There is a rapid growth in modifying AI to perform coding tasks with less to no human intervention, where models like Claude Opus 5, Gemini 3 Pro, Grok Build and Cursor agents have been operating well in recent times. 

In Codex, OpenAI has introduced a new mechanism that allows Astra to preserve and retrieve information across context windows instead of relying solely on compressed summaries. Beyond coding, Astra has been trained for scientific and mathematical applications. 

Cybersecurity is another key area of focus for the new model. OpenAI claimed that GPT-6 Astra has reached the “critical” level of cybersecurity capability under its preparedness framework, making it the company’s first model to cross this threshold.

The launch comes as OpenAI, Anthropic, Google and other AI players race to develop increasingly capable AI agents that can move beyond generating text and instead execute tasks on behalf of users.

Last year, OpenAI rolled out GPT-5 for users of ChatGPT as well as enterprises leveraging its APIs for artificial intelligence (AI) products, with enhanced features across coding, math, writing, health, visual perception, and others.

The post OpenAI Rolls Out GPT-6 Astra, Its ‘Most Capable Model’ Yet appeared first on Inc42 Media.