OpenAI and Anthropic both launched cheaper AI models on September 22, sparking an open price war between the leading AI labs. The new models arrived within hours of each other and cut running costs sharply for developers.

Anthropic released Claude Opus 5.5 at $4 per million input tokens and $20 per million output tokens, a 20 percent cut from Opus 5. Cache reads, which Anthropic says make up most of the cost in coding work, were cut 60 percent to $0.20 per million.

Hours later, OpenAI introduced GPT-6 Sol and GPT-6 Luna at roughly half the cost of the GPT-5.6 series. Sol is priced at $2 and $10 per million tokens, while Luna costs just $0.10 and $0.50, aimed at high-volume routine tasks.

xAI went the other way. Its Grok 4.7 model, released on September 21, doubles rates for prompts over 200,000 tokens, to $4 per million input and $12 per million output. Both labs are targeting the same prize: developers running long, token-hungry coding agents.

The releases are the first since the CEOs of both companies called for slowing frontier AI releases, and show competition shifting from raw intelligence to price.