September 2026 was one of the busiest months of the year in artificial intelligence: the three leading companies — Anthropic, OpenAI and Google — released new model generations within a few weeks. Here’s a summary, based on the official announcements.
In this article
Anthropic: Claude Opus 5.5
- Released on September 22, 2026, it’s the first model in the Claude 5.5 family.
- According to Anthropic, it’s stronger at coding and agentic work (long, multi-step tasks) and costs about 40% less to run than Opus 5 on typical workloads.
- Available on Claude’s Pro, Max, Team and Enterprise plans and, for developers, on Anthropic’s platform and on AWS, Google Cloud and Microsoft.
OpenAI: GPT-6 Astra, Sol and Luna
- GPT-6 Astra is the most advanced model of the new generation. OpenAI describes it as the first to reach the “Critical” cybersecurity capability level in its Preparedness Framework — so it ships with extra safeguards and a gradual rollout, first to selected organizations and then to ChatGPT Plus, Pro, Business and Enterprise plans and the API.
- GPT-6 Sol and GPT-6 Luna bring much of Astra’s progress into faster, cheaper models, with API prices 50% lower than GPT-5.6 promotional pricing, according to OpenAI.
Google: Gemini 3.8
- Gemini 3.8 Flash, released in early September, aimed at complex reasoning and software engineering. It’s in the API (Google AI Studio), Gemini Enterprise and the Google AI Pro and Ultra subscriptions.
- Google also released Gemini 3.8 Flash Cyber (security-focused), the Gemini 3.8 Live voice models and new text-to-speech models.
What it means in practice
- For everyday users: new versions roll out gradually to the apps (Claude, ChatGPT, Gemini). In most cases there’s nothing to do — the new model appears as an option or becomes the default.
- For businesses: launches come so fast that switching tools with every announcement isn’t worth it. What matters more is having clear usage rules and using business plans that protect your data — see how to use AI at work safely.
- Security: the emphasis all three companies put on cybersecurity capabilities shows AI is already used by both defenders and attackers. Updates, strong passwords and two-step verification remain the foundation.
