Three big model launches landed in the week of 21-25 September 2026: Anthropic’s Claude Opus 5.5, OpenAI’s cheaper GPT-6 Sol and GPT-6 Luna, and xAI’s Grok 4.7. For most people the practical news is price: strong models got cheaper, and a very capable model (GPT-6 Luna) is now available to free ChatGPT users on desktop.
Claude Opus 5.5 (Anthropic, 22 September)
Anthropic released Claude Opus 5.5, describing it as leading in agentic coding, computer use and knowledge work such as long financial analysis and codebase migrations. It is available through Claude, the Claude API (claude-opus-5-5) and the major clouds – AWS, Google Cloud and Microsoft Azure.
- Price: $4 per million input tokens and $20 per million output tokens – 20% lower than Opus 5 – with cheaper cache reads. Anthropic says typical workloads cost about 40% less.
- Coming next: Anthropic says Sonnet 5.5 and Haiku 5.5 are on the way.
GPT-6 Sol and GPT-6 Luna (OpenAI)
OpenAI added two cost-efficient models trained with methods similar to its flagship GPT-6 Astra. Sol balances capability and cost; Luna is built for speed and high volume.
- Price (API): Sol $2 / $10 and Luna $0.10 / $0.50 per million input / output tokens – about half the price of the GPT-5.6 versions they replace.
- Where: ChatGPT Plus, Pro, Business, Enterprise and Edu; Luna is also available to free and Go users in the desktop app; API names
gpt-6-solandgpt-6-luna.
Grok 4.7 (xAI, 21 September)
xAI released Grok 4.7, a larger base model positioned for chat and coding, at the same reported API price as Grok 4.6: $2 per million input tokens and $6 per million output tokens.
What it means for you
- If you use AI through apps: you do not need to do anything – the new models roll out inside ChatGPT, Claude and Grok. Re-run a favourite prompt or two and see whether the answers improved.
- If you build with the API: check your bills. Moving routine jobs to GPT-6 Luna or similar small models, and reserving Opus 5.5 or GPT-6 Astra for hard tasks, can cut costs sharply. Estimate it with the Token & Cost Calculator.
- If you are choosing a model: the AI models guide is updated with these releases, and the AI Model Picker gives a recommendation in three questions.
Try this after any model update
Keep three prompts you use every week and re-run them whenever your AI tool announces a new model. It takes five minutes and shows you what actually changed for your work – better than any benchmark chart. The Prompt Compare tool helps you see the differences.