Anthropic is considering releasing a new AI model to reclaim enterprise ground lost since OpenAI launched GPT-6 Astra on September 3, according to three informed sources cited by Reuters1,2. The deliberation is unfolding as Anthropic prepares for an IPO and just days after CEO Amodei publicly urged the industry to slow capability releases.
The company is still evaluating the safety of the prospective model and has not determined a final release date.
The competitive pressure is quantifiable. Reuters cited Ramp data showing GPT-6 Astra currently accounts for about 13% of enterprise AI spending, while Anthropic's Claude Fable holds around 8%. Anthropic's annualized revenue run rate exceeded $65 billion as of the end of July, with projections of $190 billion to $200 billion by 2028. OpenAI's annualized revenue run rate during the same period exceeded $40 billion.
ANALYSIS Anthropic leads OpenAI on absolute revenue run rate, yet trails on the enterprise-spending share metric that Ramp tracks, suggesting GPT-6 Astra is shifting wallet share among business buyers even as Anthropic's top line remains larger.
On September 12, Amodei published a 3,800-word article calling on the industry to slow the release of new capabilities and focus more on safety risks. Some investors have begun reassessing Anthropic's competitive position in the enterprise AI market following GPT-6 Astra's launch.
ANALYSIS A model release timed to an IPO roadshow, arriving less than two weeks after Amodei's public call for restraint, would test whether the market reads the move as pragmatic competitive response or as tension with the company's stated safety posture.
The potential launch would come shortly after Anthropic selected Accenture as its first embedded safety evaluator in a $1 billion, five-year deal to red-team models and test safeguards[2]. That arrangement could provide a framework for the safety evaluation the company says is still underway on the new model.