Anthropic is considering releasing a new AI model as it prepares for a possible IPO and responds to the enterprise momentum of OpenAI’s GPT-6 Astra, according to a September 19 Reuters report citing three sources. Anthropic has not announced the model or confirmed a launch date, and the company declined to comment for the report.
A new model could help Anthropic defend its enterprise position, but it would not settle the Anthropic IPO question by itself. Investors will compare customer adoption, revenue quality, computing costs, margins and safety controls. The company must also explain how a faster commercial response fits with CEO Dario Amodei’s call to slow the pace of frontier-model capability growth.
What Is Anthropic Considering Before Its IPO?
The Reuters report published September 19 says Anthropic is discussing whether to release another model to answer competitive pressure from OpenAI. The decision reportedly includes a safety review and a financial question: whether a launch can strengthen Anthropic’s market position without pushing profitability further away.
The wording matters. Anthropic is considering a model. It has not provided a public model name, capability list, price or release date. A responsible news article should therefore evaluate the business decision rather than describe an unannounced product as available.
The Reported Model Decision
Anthropic built its enterprise position around Claude, particularly coding, professional workflows and customers that place a high value on controlled deployment. OpenAI’s September release of GPT-6 Astra gives those customers another high-capability option.
A new Anthropic model could improve benchmark performance, reduce inference costs or add agent features. Any of those changes might help sales. The commercial value will depend on whether companies expand paid usage after testing it, not on a launch presentation or a short benchmark lead.
The Reuters sources also said Anthropic is evaluating safety as part of the release decision. That process could affect launch timing if the model shows new capabilities that require stronger controls or independent testing.
Why the Anthropic IPO Date Is Still Unsettled
Reuters reported that Anthropic could push its IPO until after the November US midterm elections. Earlier plans had already moved, with marketing previously expected to begin no earlier than mid-October.
An expected IPO is different from a filed timetable. Until Anthropic submits formal documents and announces a process, an exact Anthropic IPO date remains conditional. Market conditions, audited financial preparation, governance work and model-safety questions can all change the schedule.
OpenAI has reduced some competitive pressure to list quickly. Reuters said OpenAI CEO Sam Altman confirmed that OpenAI would not go public in 2026, citing AI-safety concerns. Anthropic can still choose a different path, but it does not need to beat OpenAI to the public market simply to preserve investor interest.
Why OpenAI Astra Increased Pressure on Anthropic
The strongest evidence of pressure is not that Astra exists. It is that enterprise-spending and developer-traffic indicators shifted after the release.
Enterprise Spending Shift
Reuters cited data from corporate expense platform Ramp showing Astra accounted for about 13% of the enterprise AI spending it tracked, compared with about 8% for Anthropic’s Claude Fable. This suggests Astra won meaningful early business usage.
The percentages do not represent the entire AI market. Ramp measures spending visible through its own platform, so its customer mix affects the result. The signal is still useful because it compares paid enterprise behavior rather than social-media interest.
If Astra’s share remains higher for several months, Anthropic may need a product response or a clearer pricing advantage. If the gap narrows after initial testing, the launch may have reflected customers experimenting with a new model rather than permanently switching providers.
Developer-Traffic Signal
OpenRouter told Reuters that users spent more on OpenAI models than Anthropic models during the measured week. It was the first OpenAI lead on that indicator in more than two and a half years.
Developer routers make it easier to move traffic among models, so their data can react faster than enterprise contracts. The shift shows immediate developer interest, but it may also be more volatile than spending from a bank, software company or large professional-services firm with a long deployment cycle.
Why Enterprise Customers Do Not Switch Overnight
Large companies integrate models into security policies, data systems and employee workflows. Changing a provider can require new testing, procurement approval, privacy review and retraining. A model that performs better on one task may still lose if migration costs are higher than the expected benefit.
Anthropic’s installed enterprise relationships therefore remain valuable. Reuters cited investors who believe this position will be difficult to displace quickly. The more important question is whether Anthropic keeps existing workloads and wins new ones as contracts renew.
Anthropic Still Has a Revenue Lead
The same Reuters report that described Astra’s momentum also showed Anthropic retaining a larger reported revenue run rate. This prevents the competition from being reduced to one week of model usage.
Revenue Run Rate vs Profitability
Anthropic’s annualized revenue run rate exceeded $65 billion at the end of July, according to Reuters, up from about $9 billion at the end of 2025. OpenAI’s annualized run rate had passed $40 billion in July.
Annualized run rate takes a recent period and projects it across a year. It shows current sales pace, but it is not the same as audited annual revenue. It also says nothing by itself about profit because frontier AI companies spend heavily on computing capacity, research and customer acquisition.
For an Anthropic IPO, investors will want to know how much revenue remains after model-serving costs. A new model that attracts customers but requires much more computing per request may improve the top line without improving the path to cash flow.
Higher Interest Rates Raise the Bar
When interest rates rise, future profits are worth less in today’s valuation models. Investors become less willing to value a company only on distant growth and more interested in when operating cash flow can cover spending.
This is especially important for AI labs because they must reserve chips, data-center capacity and power before customer revenue is fully known. Anthropic’s model decision therefore combines product competition with capital allocation: the company must decide whether the expected revenue gain justifies the additional training and deployment cost.
Can Anthropic Move Faster Without Weakening Its Safety Position?
Anthropic has marketed safety as a product and governance advantage. The company now has to show that safety can remain credible when a competitor changes the commercial race.
The Call to Pace Frontier AI
On September 12, Dario Amodei published “We Must Pace the Frontier,” arguing that frontier-model capabilities should advance slowly enough for safety work to keep pace. His proposal included stronger evaluation and coordination rather than a permanent stop to AI development.
Anthropic followed that message with a concrete step. On September 18, the company announced a partnership with Accenture on embedded evaluation, describing it as independent evaluation of frontier AI and a commitment connected to Amodei’s essay.
This gives Anthropic a way to argue that a new launch and safety scrutiny can happen together. The test will be whether independent evaluators receive meaningful access and whether material findings change deployment decisions.
The Commercial Pressure
If Anthropic delays too long, developers may build new workflows around Astra and become harder to win back. If it releases too quickly, any safety or reliability failure would damage the position that differentiates Claude.
The practical middle path is a staged rollout: limited access, independent evaluation, monitored enterprise deployment and broader release only after identified issues are addressed. This approach may look slower than a full public launch but can create stronger evidence for customers and IPO investors.
What Could Affect an Anthropic IPO Valuation?
Four measurable factors will matter more than a single model announcement.
Enterprise Customer Retention
Investors will examine how much existing spending remains after Astra reaches more companies. Renewal rates, usage expansion and multi-year agreements can show whether Anthropic’s enterprise lead is durable.
Model Usage Growth
API traffic, paid seats and the mix of high-value professional tasks provide a clearer demand signal than download or social metrics. Growth across several customer groups is stronger than dependence on one large account.
Margin Direction
Training costs receive attention, but inference costs are paid every time a customer uses the model. Anthropic can improve margins through more efficient models, higher utilization and pricing that reflects the value of complex tasks.
Customer Concentration and Open-Source Competition
Reuters reported that Meta, one of Anthropic’s largest customers, is seeking to reduce its reliance on Anthropic as it develops more internal AI capability. A large customer bringing work in-house can affect both growth and investor confidence.
Open-source and open-weight models create another form of pressure. They allow companies to host more of their own infrastructure and negotiate harder on API pricing. Anthropic must compete with OpenAI while also proving that a paid frontier model provides enough reliability, support and performance to justify its cost.
Tapbit Learn’s article on OpenAI and Anthropic Pre-IPO perpetuals explains how market contracts differ from private shares. The overview of OpenAI’s AI infrastructure spending adds context on why computing commitments affect valuation.
Anthropic vs OpenAI: What to Watch Next
| Measure | Anthropic | OpenAI | Why It Matters |
|---|---|---|---|
| Reported July annualized revenue run rate | Above $65 billion | Above $40 billion | Shows current sales pace, not profit |
| Ramp enterprise-spend signal | Claude Fable about 8% | Astra about 13% | Measures tracked paid business usage |
| Next model status | Under consideration, per Reuters | GPT-6 Astra released | Tests product response and adoption |
| 2026 IPO status | Timing may move beyond November | No 2026 IPO, per Reuters | Changes investor access and disclosure timing |
| Safety approach | Embedded external evaluation announced | Third-party evaluation commitments reported | Affects trust and release speed |
The next decisive evidence will be sustained usage rather than another headline. Watch whether Ramp and OpenRouter signals persist, whether Anthropic announces a model and whether the company provides a formal IPO filing or timetable.
Tapbit Learn’s coverage of Nvidia and Hugging Face also shows how infrastructure and open-model ecosystems influence the competitive field around frontier labs.
How to Trade ANTHROPIC-USDT or OPENAI-USDT on Tapbit
Tapbit lists ANTHROPIC-USDT and OPENAI-USDT as Pre-IPO-related perpetual futures markets. They provide price exposure through contracts and do not represent private-company shares, shareholder voting rights or a guaranteed allocation in a future IPO.
- Register a Tapbit account or log in.
- Open the ANTHROPIC-USDT or OPENAI-USDT futures page.
- Confirm the selected Pre-IPO-related contract and review its market information.
- Choose a direction and order type, then enter the intended position size.
- Review margin requirements and the full contract order before confirming.
- Monitor the position around model announcements, enterprise-usage reports and IPO updates.
Frequently Asked Questions
Is Anthropic releasing a new AI model?
Reuters reported on September 19 that Anthropic is considering a new model, citing three sources. Anthropic has not publicly confirmed the model or a release date.
When is the Anthropic IPO?
No final Anthropic IPO date has been announced. Reuters reported that the offering could move until after the November 2026 US midterm elections.
Is Anthropic ahead of OpenAI in enterprise AI?
The answer depends on the measure. Anthropic had the higher reported July annualized revenue run rate, while Astra led Claude Fable in the Ramp enterprise-spending snapshot cited by Reuters.
What is Anthropic’s reported revenue run rate?
Reuters reported that Anthropic’s annualized revenue run rate exceeded $65 billion at the end of July 2026. Run rate is a projected sales pace, not audited annual revenue or profit.
Does ANTHROPIC-USDT represent Anthropic shares?
No. ANTHROPIC-USDT is a Pre-IPO-related perpetual futures contract on Tapbit. It does not provide ownership of Anthropic, voting rights or shares in a future IPO.
