Anthropic is weighing whether to release a new AI model in the coming months, according to three people familiar with the matter cited by Reuters, as the company prepares for an anticipated initial public offering and watches OpenAI’s GPT-6 Astra pick up enterprise customers. The timing matters because it puts commercial pressure squarely against the safety-first message Anthropic has built its brand around.
The company is currently evaluating the safety profile of its next model while its leadership debates a separate, related question: whether to keep pouring resources into new model development or shift emphasis toward profitability ahead of a public listing. Enterprise spending data shows why that debate has urgency. Corporate expense platform Ramp found that Astra, released September 3, already accounts for roughly 13% of enterprise AI spending it tracks, compared with about 8% for Anthropic’s Claude Fable.
The deliberations follow closely on the heels of a public warning from Anthropic’s own CEO about the pace of AI development, meaning any near-term launch will draw scrutiny over whether the company is following its own advice. How Anthropic resolves that tension, and what its next model looks like if it ships, will likely shape how investors price the company’s IPO.
A Slowdown Call, Then a Possible Launch
Just over a week before the Reuters report, Anthropic CEO Dario Amodei published a lengthy essay arguing that “we must slow the pace at which we improve the capabilities of AI models,” warning that unchecked progress could let swarms of AI agents effectively take over the internet within six to twelve months. The essay drew public support from OpenAI CEO Sam Altman and SpaceX CEO Elon Musk, an unusual show of agreement among rival executives on AI safety. Investing.com
The Optics Problem
That essay is now the backdrop against which any new Anthropic model launch will be read. A company whose chief executive just called for industry restraint would be moving to ship a more capable product within weeks, a contradiction that Anthropic’s leadership appears aware of as it weighs the decision, according to the sourcing behind the Reuters report.
GPT-6 Astra’s Enterprise Push
OpenAI has framed Astra, launched September 3, as its strongest model yet for enterprise use cases including software engineering, cybersecurity and general professional work. The early numbers suggest that positioning is resonating with business customers who had previously defaulted to Anthropic’s models.
Where the Numbers Stand
| Metric | OpenAI (GPT-6 Astra) | Anthropic (Claude Fable) |
|---|---|---|
| Share of enterprise AI spending (Ramp) | ~13% | ~8% |
| Developer spend on OpenRouter, past week | Higher, for first time in 2.5+ years | Lower |
Beyond the Ramp figures, developer routing platform OpenRouter reported that its users spent more on OpenAI’s models than Anthropic’s last week, the first time that has happened in more than two and a half years. That shift, more than the raw percentages, is what appears to have investors asking questions ahead of Anthropic’s IPO.
Background: Anthropic’s Enterprise Standing
Anthropic has spent much of the past year positioned as the default choice for enterprise AI deployments, a reputation built on Claude’s coding and reasoning performance. Some investors reportedly still view that position as durable despite Astra’s early traction, arguing that enterprise relationships and switching costs favor incumbents.
What Anthropic Is Actually Deciding
The choice in front of Anthropic is not simply whether to ship a new model. It is whether to prioritize aggressive capability gains to defend market share, or to slow down, as Amodei’s own essay urged, while leaning harder into profitability with existing products ahead of a public offering. Anthropic did not respond to Reuters’ request for comment on the deliberations.
The initial report treats these discussions as ongoing rather than settled, meaning no launch date, model name, or final decision has been confirmed as of this writing.
