The artificial intelligence arms race may be accelerating again.
Anthropic is considering releasing a new AI model as it responds to the early momentum surrounding OpenAI’s GPT-6 Astra, according to a Reuters report citing three people familiar with the company’s plans. No final decision has reportedly been made on whether—or when—the model will launch.
The discussions come at a pivotal moment for Anthropic. The Claude maker is reportedly preparing for a potential initial public offering while simultaneously trying to protect its position in the lucrative enterprise AI market.
And GPT-6 Astra has added a new variable to that equation.
GPT-6 Astra Is Putting Pressure on the Competition
OpenAI officially introduced GPT-6 Astra on September 3, calling it its most capable broadly deployed model yet. The company says Astra represents a major advance across computer use, software engineering, browsing, science and professional work.
Perhaps more important for Anthropic, Astra appears to be gaining traction with businesses.
Reuters reported that Astra has recently taken the lead in spending through OpenRouter and has been gaining ground in enterprise AI budgets. Other reports citing Ramp spending data put Astra at roughly 13% of tracked enterprise AI spending, compared with approximately 8% for Anthropic’s Claude Fable.
Those figures represent a snapshot of spending measured through particular platforms rather than the entire AI market. Still, they offer an indication of how quickly companies are experimenting with the latest frontier models.
That matters because enterprise adoption has become one of the biggest battlegrounds in AI.
Anthropic Has a Major Decision to Make
Anthropic isn’t simply deciding whether it has another model ready.
The company must decide how aggressively it wants to compete in an industry where frontier models can require enormous amounts of computing infrastructure and capital.
Reuters reported that Anthropic is weighing development expenses, profitability and safety considerations as it evaluates a potential release. The company could also face increased scrutiny from investors as it considers an IPO.
That creates an interesting tension.
Anthropic CEO Dario Amodei has recently been among the AI leaders raising concerns about the speed at which increasingly powerful systems are developing. Yet competitive pressure from OpenAI and other laboratories continues to push the industry toward faster development cycles.
Now Anthropic may have to balance those two realities.
AI Is Already Helping Anthropic Build Its Next AI
There is another remarkable part of this story.
Anthropic recently disclosed that Claude itself is increasingly involved in developing future versions of the technology.
According to reporting by the Associated Press, Anthropic said Claude accounted for about 26% of its model research and development work under human supervision as of August, up from essentially zero in February. The AI reportedly participates in some capacity in roughly 90% of the company’s R&D tasks.
That doesn’t mean Claude is independently creating its successor. Human researchers remain involved and supervise the work.
However, the trend points toward one of the most consequential developments in artificial intelligence: AI systems increasingly assisting researchers who are building the next generation of AI systems.
If that process becomes dramatically more capable, development cycles could potentially accelerate.
OpenAI Has Made Astra About More Than Chat
GPT-6 Astra also illustrates how the competition is moving beyond traditional chatbots.
OpenAI says Astra can operate computers, browse the web, work inside software environments and perform complex professional tasks. The company reports significant improvements in computer-use performance compared with GPT-5.6 Sol.
Astra also reached OpenAI’s “Critical” cybersecurity capability threshold under its Preparedness Framework. OpenAI says that designation means the model can, given appropriate tools and access, identify previously unknown vulnerabilities and develop methods for exploiting protected systems without requiring human direction at every step. The company says it introduced additional safeguards before deployment because of those capabilities.
That combination of capability and risk helps explain why the next stage of the AI competition is becoming increasingly complicated.
Companies aren’t simply racing to build smarter chatbots anymore.
They are developing systems capable of taking actions.
The Enterprise AI Battle Is Becoming Critical
Anthropic has built much of Claude’s reputation around coding, enterprise applications and safety. OpenAI, meanwhile, is increasingly positioning Astra as a platform capable of performing substantial amounts of professional computer-based work.
The distinction matters.
The biggest economic opportunity for generative AI may eventually come from businesses delegating entire workflows to AI rather than employees simply asking chatbots questions.
An AI system capable of researching information, operating software, writing code, analyzing data and completing multi-step assignments could function more like a digital worker than a conventional assistant.
That possibility is driving enormous investment across the industry.
It is also raising the stakes for companies such as Anthropic and OpenAI.
Safety and Speed Are Colliding
The timing of Anthropic’s reported deliberations highlights a fundamental challenge facing the entire AI industry.
Leading researchers continue to warn about increasingly capable AI systems while the companies employing those researchers face intense pressure to release increasingly capable AI systems.
OpenAI itself recently disclosed examples of concerning behavior observed during AI testing and introduced a framework designed to better track potential model misalignment.
Anthropic has similarly made safety research central to its identity.
Yet neither company operates in isolation.
Open-source models, major technology companies and international competitors continue improving their own systems. That makes slowing development difficult even when executives believe additional caution is warranted.
What Happens Next
For now, Anthropic has not publicly announced the rumored model or provided a launch date.
That distinction is important: reports that the company is considering a release should not be interpreted as confirmation that a new Claude model is imminent.
Still, the competitive landscape has changed quickly since GPT-6 Astra arrived.
OpenAI has pushed frontier AI further into autonomous computer use and professional workflows. Anthropic remains one of its strongest competitors, particularly among businesses and developers.
If Anthropic does introduce another major Claude model soon, the next phase of the AI race could arrive much faster than expected.
And increasingly, the systems competing in that race may also be helping engineers build whatever comes next.
Sources: Reuters, Associated Press and OpenAI.
