OpenAI has launched a new AI model, GPT-6 Astra. The company calls it the smartest in the world

OpenAI, one of the leading artificial intelligence startups, has launched its new AI model, GPT-6 Astra, which it calls “the smartest in the world,” according to the Financial Times. According to the company, Astra leads the way in software development, science, and cybersecurity, and outperforms even the most advanced AI models from its competitor, Anthropic.

GPT-6 became available on Thursday, September 3, to business users participating in the Daybreak program, which allows verified testers to use OpenAI’s most advanced models, according to Bloomberg. Astra is not yet available to the general public, but the developer says it plans to release it in the coming days. It plans to release a version of the model for paid ChatGPT users with stricter restrictions to ensure cybersecurity, Bloomberg reports. However, the agency added that limits preventing users from taking advantage of the model’s most innovative capabilities in this area are also built into the main version.

"Astra includes additional safety checks in case agents might have misinterpreted your instructions. If a potential issue of this kind is detected, the conversation will be paused or stopped as a safety measure so you can assess the situation and decide how to proceed,” according to a statement from OpenAI.

Using Astra will cost about the same as OpenAI’s main competitor, Anthropic’s flagship model, Claude Fable 5.1, according to the FT. At the same time, the publication notes that the adoption rate of Anthropic’s top model has stopped growing since its launch, as users are switching to cheaper alternatives such as Claude Opus 5.

OpenAI stated that Astra will be more effective than previous-generation models. The developer claims that Astra demonstrates particularly strong results in financial modeling, outperforming humans in the Financial Modeling World Cup, as well as in preparing tax returns and analyzing data. In addition, the model handles “routine tasks,” such as filling out forms, quite well.

OpenAI President Greg Brockman said at a press briefing on Thursday that the new model “represents a generational leap in capabilities” and that it can be considered general artificial intelligence (AGI), according to the FT. This term refers to the stage at which AI outperforms humans across a broad range of cognitive tasks, the publication explains.

“Everyone has their own definition of AGI. It’s a vague, ambiguous concept. But I think that when we look back, people will come to see that it happened around this time and was linked specifically to this model,” Brockman asserts.

OpenAI had previously presented AGI as a specific milestone in the development of artificial intelligence and had even included provisions regarding AGI in its multibillion-dollar investment agreements with Microsoft and Amazon. However, on Thursday, Brockman stated that AGI now represents “more of a mission statement or a guiding principle,” according to the FT.

OpenAI led the market following the launch of ChatGPT in late 2022. However, this year the company has ceded the lead to Anthropic, according to the FT. Anthropic’s valuation has risen to $965 billion ahead of its IPO, which is expected to value the company at roughly twice that amount this year, the publication notes. OpenAI’s valuation stood at $852 billion following its funding round in March.

This article was AI-translated and verified by a human editor