Complex reasoning and professional work
GPT-6 Astra is described as OpenAI's most capable model for complex reasoning, science, professional work, software engineering, and computer use.
Coding / product dossier
GPT-6 Astra
OpenAI model for multistep agents, reasoning, coding, computer use, and science.
Product brief
What GPT-6 Astra does.
GPT-6 Astra is OpenAI's most capable model for complex reasoning, software engineering, computer use, science, and professional work. Async tool calling and mid-turn steering help it run multistep agent workflows. API: model id gpt-6-astra, $10/$50 per 1M tokens (short context). Rolling out first via Trusted Access/Daybreak, then Plus, Pro, Business, Enterprise, and the API in the coming days. First OpenAI model at the Critical cybersecurity capability threshold (advanced cyber access limited).
Why we selected it
A consequential new frontier model with concrete agent-workflow features, plus direct relevance to this week’s Copilot and workplace rollout news.
- Best for
- Teams building high-capability agents
- Category
- Coding
- Daily picks
- 1
- First selected
- 2026-09-04
Product preview saved with our daily selection
Capability scan
What it can help with.
Only capabilities supported by the product information we collected are listed here.
Multistep agent workflows
Async tool calling and mid-turn steering are intended to help the model run multistep agent workflows.
API access
The stated API model ID is gpt-6-astra.
Advanced cyber access restrictions
The product is described as OpenAI's first model at the Critical cybersecurity capability threshold, with advanced cyber access limited.
Best-fit use cases
Build agents that carry out multistep work with tool calls.
Support software-engineering workflows.
Use a model for complex reasoning in scientific or professional tasks.
Develop high-capability agents through the gpt-6-astra API model.
FAQ
Before you open it.
What is GPT-6 Astra designed for?
It is described for complex reasoning, software engineering, computer use, science, professional work, and multistep agent workflows.
What features support agent workflows?
The description names async tool calling and mid-turn steering.
What API model ID should developers use?
The listed model ID is gpt-6-astra.
What API pricing is stated?
The stated short-context price is $10 per 1 million input tokens and $50 per 1 million output tokens.
How is access rolling out?
It is stated to roll out first through Trusted Access/Daybreak, followed by Plus, Pro, Business, Enterprise, and the API in the coming days.