Anthropic holds new model under pre-IPO pressure and GPT-6 advancement Astra

Anthropic is assessing the safety of its next model before deciding whether and when it will launch it, according to sources heard by Reuters. The pressure is double: investors follow the preparations for a possible initial public offer (IPO), while GPT-6 Astra, launched at the beginning of the month, gains space among corporate customers and puts Anthropic's leadership in AI in check for companies.

Why is Anthropic holding the launch of the new model?

Security assessment is not a marketing step: it works as a gate. According to the calculation published by PYMNTS based on Reuters sources, the company has not yet decided whether the model will be released, which leaves open the possibility of delay or launch with restricted capabilities. The movement dialogues with the attitude that Anthropic has been trying to consolidate in the market, as in the case of the cybersecurity model that surpassed humans in tests and came along with the promise to debate risks openly.

What the GPT-6 Astra changes in the corporate dispute of AI

Anthropic's brake doesn't happen in a vacuum. The GPT-6 Astra reached the market at the beginning of the month and, according to the same calculation, has been gaining corporate customers — exactly the land where Anthropic positioned itself as a preferred supplier. In AI for companies, contracts are rarely exchanged overnight: integrations, historical data and layers of compliance create inertia. But racers won by a competitor tend to become large-scale contracts months later. It is this window that Anthropic tries not to lose, in a scenario where the arms race between OpenAI, DeepSeek and Anthropic itself continues accelerated.

Does the pre-IPO pressure accelerate or stop the Anthropic launch?

Investors preparing for a capital opening want two things at the same time: revenue predictability and a product pipeline that supports valuation. A model retained by security assessments protects the company from bad headlines, but weakens the growth narrative before competition. The dilemma is the same that crosses the AI IPOs wave with valuations in the trillions’ home — and persistent bubble doubt. The difference here is that the brake is technical, not financial.

What changes to Brazilian companies that buy AI from Anthropic

For technology teams in Brazil, the practical sign is instability in the roadmap of suppliers. Who bet on a single model for internal products or customers runs the risk of seeing deadlines slip by decisions made thousands of kilometers away. The safest reading is to treat models as replaceable components: maintain layers of abstraction, avoid deep proprietary dependencies and reserve evaluation windows before migrating critical service contracts, data analysis or internal automation.

What is not yet known about the new Anthropic model

The calculation did not inform launch date, benchmarks, prices or model architecture under review. There is also no official confirmation from Anthropic about the stage of the evaluations or about the release format — broad release, limited prior release or invitation access. Without this data, any projection on performance or commercial impact is speculation. What is put is the calendar: the GPT-6 Astra is already in the market, and Anthropic is still at the exit gate.

Frequently Asked Questions about Anthropic, pre-IPO and GPT-6 Astra

Has Anthropic decided to launch the new model yet?

Nope. According to sources heard by Reuters, the company still evaluates the safety of the model and did not hit the hammer on the launch.

Why does GPT-6 Astra press Anthropic?

Because the model, launched at the beginning of the month, has been gaining ground among corporate customers — precisely the segment in which Anthropic built its leadership.

What do pre-IPO investors expect from Anthropic?

Predictability of revenue and a consistent product pipeline, two items that become more fragile when an important release goes into standby.