OpenAI Releases GPT-6 Astra for Coding and Computer Use

OpenAI released GPT-6 Astra, a model focused on computer use, coding, professional workflows, science, and cybersecurity, initially available to a limited set of organizations and rolling out to ChatGPT Plus, Pro, Business, and Enterprise users plus the OpenAI API, Microsoft Azure, and AWS Bedrock. OpenAI reported Astra scored 72.6% on OSWorld 2.0 versus 65.7% for GPT-5.6 Sol, 57.9% on Terminal-Bench 4.0, 74.1% on DeepSWE v1.1, and 96.3% on MRCR long-context evaluations in the 512K-to-1M range, while hallucination rates fell to 4.2% from 12.2%. Astra is the first OpenAI model classified at the critical cybersecurity capability level under the company's Preparedness Framework, and in testing without production safeguards OpenAI said it discovered and used two previously unknown vulnerabilities and developed exploits against hardened browsers and operating systems.

OpenAI has released GPT-6 Astra https://openai.com/index/gpt-6-astra/ , a new model focused on computer use, coding, professional workflows, science, and cybersecurity. The model is initially available to a limited set of organizations and is rolling out to ChatGPT Plus, Pro, Business, and Enterprise users, as well as the OpenAI API https://openai.com/api/ , Microsoft Azure, and AWS Bedrock. Astra extends OpenAI's models beyond generating responses toward performing multi-step tasks directly in software. It can interact with graphical interfaces to fill forms, update CRM records, conduct research, create websites, analyze data, install and test software, and troubleshoot problems visible on screen. On OSWorld 2.0 https://benchlm.ai/benchmarks/osworld2 , OpenAI reports a score of 72.6%, compared with 65.7% for GPT-5.6 Sol. Coding is another focus of the release. OpenAI reports 57.9% on Terminal-Bench 4.0 https://benchmarklist.com/benchmarks/terminal bench 4/ and 74.1% on DeepSWE v1.1. Astra also introduces an experimental context https://learn.chatgpt.com/docs/config-file/config-reference mechanism in Codex https://openai.com/codex/ that allows the agent to maintain notes across context windows instead of relying only on compaction. Previous context windows remain searchable, allowing the model to retrieve earlier requirements, test results, and tool outputs during long-running coding tasks. The model supports long contexts of up to one million tokens in OpenAI's reported MRCR evaluations https://openai.com/index/gpt-6-astra/ :~:text=%2D-,Long%20Context,-Long%20Context , scoring 96.3% in the 512K-to-1M range. OpenAI also reports improvements in professional tasks including database migrations, CAD generation, data science, browser research, and scientific workflows. Cybersecurity represents a significant change from previous OpenAI models. Astra is the first OpenAI model classified at the critical cybersecurity capability level under the company's Preparedness Framework https://openai.com/index/path-to-astra/ . In testing without production safeguards, OpenAI says the model discovered and used two previously unknown vulnerabilities and demonstrated the ability to develop exploits against hardened browsers and operating systems. The production version restricts advanced offensive tasks, while OpenAI plans to provide broader defensive capabilities through its Daybreak https://openai.com/daybreak/ program. OpenAI also reports lower hallucination rates in its internal evaluation, with Astra scoring 4.2% compared with 12.2% for GPT-5.6 Sol. However, the company found Astra's written reasoning harder to monitor than its predecessor in tests designed to measure whether a model could obscure its reasoning. OpenAI said improving this monitorability remains an active research area. Community reaction included comments from Nvidia CEO Jensen Huang https://x.com/JensenHuang/status/2096700264569090384 , who highlighted the infrastructure used to train the model, saying: GPT-6 Astra, trained on ~100K+ NVIDIA Grace Blackwell NVLink72. From ChatGPT to o1 to Astra in 4 years. AGI has arrived. Congratulations @OpenAI team. 400K GPUs coming online next. Another community voice, Alex Finn https://x.com/AlexFinn/status/2095581065884909661 , similarly focused on the AGI framing around the release, pointing to the characterization of the current period sharing: Welcome to AGI. ChatGPT 6 Astra just released. Astra competes with models including Anthropic's Claude Fable 5.1 https://www.infoq.com/news/2026/06/claude-5-release/ and Google's Gemini 3.8 Flash. OpenAI's published evaluations show results varying by task: Astra leads the compared models on Terminal-Bench 4.0 https://benchmarklist.com/benchmarks/terminal bench 4/ and several computer-use evaluations, while Claude Fable 5.1 scores higher on Humanity's Last Exam and Google's Gemini 3.8 Flash supports native video and audio input that Astra does not.