Anthropic has filed for an IPO, with its confidential prospectus revealing a 2025 revenue of $4.6 billion against an operating loss exceeding $8 billion. By the second quarter of 2026, revenue reached $11.5 billion, marking the company's first positive adjusted operating profit.
The company also released Sonnet 5.5, which offers up to 30% cost savings per task due to lower token consumption. Additionally, Anthropic analyzed the Chinese GLM-5.3 model, noting its ability to identify vulnerabilities and assemble exploit chains comparable to proprietary models.
OpenAI introduced 'Dots' agents, capable of executing tasks autonomously across applications. The company also updated its subscription tiers, introducing a $500 Pro plan while halving usage limits for new subscribers on the $200 tier, effective October 29.
OpenAI delayed the release of GPT-6.1 Astra due to safety concerns, opting to launch the more affordable GPT 6.1 Sol instead. Similarly, Google has restricted access to its new Gemini 4 Argon model to cybersecurity specialists for further safety testing.
President Trump and six major AI companies, including Google, OpenAI, Anthropic, Meta, Nvidia, and xAI, signed a voluntary safety pact. The agreement mandates internal risk monitoring and independent audits, though it carries no legal penalties for non-compliance.
