Anthropic Weighs a New Claude Model as OpenAI's Astra Overtakes It in Tracked Enterprise Spending

Anthropic is considering a new model release and is running safety evaluations on it, but has not fixed a launch date, Reuters reported. The trigger is commercial. OpenAI shipped GPT-6 Astra on September 3 with claimed gains in computer use, software engineering and cybersecurity work, and enterprise spending began moving within two weeks.

The numbers cited in the reporting come from third-party trackers rather than either company. Ramp's corporate-card data puts Astra at roughly 13% of tracked enterprise AI spending against about 8% for Claude Fable, and OpenRouter reported that OpenAI model spending passed Anthropic's for the first time in about two and a half years, as summarized by AI Post. Both measures capture a narrow slice of the market. Ramp sees card transactions, not negotiated enterprise contracts, and OpenRouter sees routed API traffic, not direct commitments. Anthropic's revenue base is larger than either proxy implies.

The timing matters more than the share figures. Anthropic filed confidentially for an initial public offering on June 1 and investors have discussed an October debut at $2 trillion or above, a level reports describe as the largest public-market debut on record. That valuation rests on an annualized revenue run rate reported at about $65 billion at the end of July, up from roughly $9 billion at the end of 2025, and on projections of $100 billion to $120 billion by year-end. A visible loss of share to a direct competitor in the weeks before pricing is the kind of fact underwriters have to answer for.

There is an obvious tension. Anthropic chief executive Dario Amodei published an essay on September 12 arguing the industry should slow the rate at which the most capable systems improve. Eight days later the company is reported to be preparing a counter-release. Anthropic's current flagship pair, Claude Fable 5.1 and Claude Mythos 5.1, shipped on September 1, two days before Astra.