OpenAI Says Astra Beats Anthropic. Read the Caveat Underneath.
The model improves software, science and finance tasks and adds restricted cybersecurity tools that can find unknown system flaws, OpenAI said.
- On Thursday, OpenAI launched GPT-6 Astra, claiming the model reclaims the industry lead from rival Anthropic after losing ground in the race to build the most capable artificial intelligence.
- Anthropic's Claude models gained ground among professional users, prompting OpenAI to position Astra as superior for software engineering, financial modeling, and scientific work.
- During a Wednesday briefing, OpenAI President Greg Brockman claimed Astra represents artificial general intelligence . Testing by the ARC Prize Foundation confirmed the model scored 96 per cent on the ARC-AGI-3 benchmark, reaching human parity.
- OpenAI launched the model to organizations on 3 September, with rollout to ChatGPT Plus, Pro, Business, and Enterprise subscribers expected in coming days. The company restricted powerful cyber capabilities to trusted testers.
- As OpenAI and Anthropic prepare for potential stock-market listings, competition for technical leadership intensifies. Improving monitorability remains a research priority as Astra demonstrates advanced autonomy capabilities.
37 Articles
37 Articles
US-based artificial intelligence company OpenAI has announced that its newly released model, "GPT-6 Astra," is able to find high-level security vulnerabilities.
OpenAI says Astra beats Anthropic. Read the caveat underneath.
OpenAI released GPT-6 Astra on 3 September, saying it outperforms every rival including Anthropic’s Claude and Google’s Gemini. Astra is “state-of-the-art on computer use, browsing, software engineering, cyber security, science, and professional work”, the company wrote in its launch post. The Financial Times reported the launch as an attempt to retake the technical lead from Anthropic, which […] This story continues at The Next Web
The GPT-6 Astra arrived this Thursday, but not for everyone at the same time. OpenAI scaled the release of the model: it starts with a restricted group of cybersecurity organizations and only n
GPT-6 Astra: OpenAI unveils AI model that can operate computers
By Juliet Umeh OpenAI has unveiled GPT-6 Astra, a new artificial intelligence model capable of operating computers and executing complex tasks for users, in what experts describe as another major step towards autonomous AI. The model’s unveiling comes amid intensifying competition between OpenAI and rival AI company, Anthropic, with the rivalry driving rapid advances in artificial intelligence. A demonstration of Astra showed the model performin…
At the moment, it is only available to users of the Daybreak cybersecurity platform.
Coverage Details
Bias Distribution
- 53% of the sources lean Left
Factuality
To view factuality data please Upgrade to Premium
