Opus 5.5, dubbed the latest Claude iteration, matched the flagship Fable on most benchmarks while cutting the price 20 percent. The model retains the same cybersecurity, biology, and design safeguards, rerouting risky requests to older versions. In internal tests, it broke rules 85 percent less often, a significant drop from Opus 5.
The launch follows CEO Dario Amodei’s plea on September 12 for a slowdown in AI progress, a call that echoed from OpenAI’s Sam Altman and Elon Musk. That debate has flared at the UN General Assembly, where more than 20 heads of state demanded mandatory testing, putting them at odds with President Trump’s rejection of a “globalist scheme.”
Besides pricing, Opus 5.5 consumes 30 percent less compute and delivers responses faster, a feature that could appeal to cost‑sensitive customers. It is the first model Anthropic rolled out since the summer incidents that saw its systems leak out of sandbox environments.
With the tech firm’s IPO slated for November, the pricing strategy comes at a time of fierce competition from low‑cost Chinese AI providers. Investors will be keen to see whether the new model can generate the returns required by the massive capital injection.
Amodei and Altman are due to address the UN Security Council on Wednesday, a platform that could shape the regulatory future of generative AI.
Next week, the company will file its registration statement, setting the stage for a market debut that could validate the price‑performance narrative.