Anthropic Launches Faster Sonnet 5.5 as IPO Prospectus Details AI Risks and Soaring Compute Costs
aiinsider ·
Anthropic has released Claude Sonnet 5.5, a faster and more cost-efficient version of its mid-tier model designed for everyday tasks like coding and document creation. The launch coincides with the filing of its IPO prospectus, which dedicates nearly a third of its content to risk factors, including warnings about dangerous model behaviors such as resisting shutdown or manipulating information. Financial disclosures reveal that while revenue surged twelvefold to nearly $4.6 billion in 2025, the company recorded an operating loss exceeding $8 billion due to soaring compute costs. Anthropic plans to invest $518 billion in cloud and infrastructure following deals with Google, SpaceX, and Nscale, with backers projecting a listing valuation potentially exceeding $2 trillion. CEO Dario Amodei has emphasized AI safety concerns, describing artificial intelligence as a critical global security threat during recent addresses to the UN Security Council. This move follows similar releases from competitors like OpenAI and Meta, highlighting intense competition in the agentic AI space. The new Sonnet model outperforms Opus 5.5 in specific agentic coding benchmarks by efficiently spawning multiple agents within cost limits.
aiinsider