Anthropic is weighing an accelerated launch of a new AI model to blunt enterprise momentum gained by OpenAI’s recently deployed GPT-6 Astra, according to internal deliberations reported by Reuters.
Key points
- Anthropic is evaluating an accelerated release of a new AI model to counter OpenAI's GPT-6 Astra.
- Data from corporate spend tracker Ramp shows GPT-6 Astra took 13 percent of enterprise AI spending.
- OpenAI passed Anthropic on developer routing hub OpenRouter for the first time in over two years.
- The launch debate follows an essay by Anthropic CEO Dario Amodei urging the industry to slow AI capability scaling.
The debate arrives at a critical juncture for both companies. Anthropic is preparing for a potential initial public offering as early as late 2026, while enterprise clients show early signs of shifting budgets toward OpenAI’s latest software.
Enterprise Spending Shifts Toward GPT-6 Astra
OpenAI introduced GPT-6 Astra on September 3, positioning the system for professional workflow automation, computer control, cybersecurity and software engineering. The model offers an input context window between 1.05 million and 1.1 million tokens, priced at $10 per million input tokens and $50 per million output tokens.
Market uptake followed swiftly. According to data from corporate expense tracker Ramp, GPT-6 Astra captured approximately 13% of enterprise AI spending within weeks of launch. That share moved past Anthropic’s Claude Fable, which registered at roughly 8% of enterprise spending over the same tracking interval.
Developer platform OpenRouter also registered higher total spending on OpenAI models than on Anthropic systems during mid-September. The shift marks the first time in more than two and a half years that OpenAI has held the top position on OpenRouter’s developer benchmark.
Tension Between Commercial Goals and AI Safety Messaging
The prospect of a fast-tracked release follows public calls from Anthropic leadership to restrict the pace of frontier model deployment. On September 12, Anthropic CEO Dario Amodei published a 3,800-word essay titled ‘We Must Pace the Frontier,’ urging developers to slow capability advancements to evaluate catastrophic and systemic safety risks.
We must slow the pace at which we improve the capabilities of AI models.
OpenAI CEO Sam Altman publicly endorsed that position on social media, writing that OpenAI had engaged in similar internal discussions. Even as executive leadership called for restraint, Anthropic faced political scrutiny over its safety positions while navigating the financial pressures of commercial software delivery.
According to reports on Anthropic, company officials have not finalized whether to approve the new model for commercial availability. Internal review teams are actively weighing the safety validation requirements against the commercial necessity of preserving corporate accounts.
Financial Scale and Timeline Ahead of Public Listing
Despite recent pressure from GPT-6 Astra, Anthropic retains a substantial annual revenue lead over its primary competitor. Anthropic reached an annualized revenue run rate of approximately $65 billion by late July, an increase from $9 billion recorded in late 2025. Company projections anticipate run rates reaching between $190 billion and $200 billion by 2028.
| Metric | Anthropic | OpenAI |
|---|---|---|
| Annualized Run Rate (July 2026) | $65 billion | $40 billion |
| Enterprise AI Spend Share (Ramp) | ~8% (Claude Fable) | ~13% (GPT-6 Astra) |
| Flagship Model Pricing (Input/Output per 1M) | $10 / $50 (Claude Fable 5.1) | $10 / $50 (GPT-6 Astra) |
| Public Offering Timeline | Targeting late 2026 | Postponed beyond 2026 |
OpenAI disclosed an annualized revenue run rate passing $40 billion in July. Altman has confirmed that OpenAI will not pursue a public listing in 2026, citing internal safety oversight priorities. The decision leaves Anthropic alone among frontier developers facing public market timelines this year.
Broader Enterprise Pressures and Hardware Expansion
Frontier model makers face additional operational hurdles beyond competitor updates. Rising interest rates have increased investor demands for capital efficiency, while open-source alternatives continue to pressure API pricing margins. Large corporate customers, including Meta, have also expanded internal infrastructure to lower their dependence on external model providers.
Anthropic has broadened its research footprint to support broader automated workloads. In addition to standard software APIs, Anthropic opened a physical laboratory in the San Francisco Bay Area to evaluate how frontier models interact with biological hardware and laboratory equipment.
Anthropic has not publicly detailed the specifications, target release window, or formal product tier of the potential counter-model. Decisions regarding safety sign-offs and model readiness remain under internal review.
