Subscribers to professional and business plans now have access to a new suite of high-efficiency tools within the Codex and ChatGPT Work environments. These models, named GPT-6 Sol and Luna, are priced 50% lower than previous promotional rates for the 5.6 generation, with Sol costing $2 per million input tokens and $10 per million output tokens. Luna is available at $0.10 and $0.50 respectively, while Free and Go users can access the Luna model via a desktop app; according to OpenAI, Sol outperforms Claude Opus 5 on AutomationBench at roughly 9% of the cost per task.

The update introduces a Prompt Caching Dashboard and new diagnostics to help developers reduce redundant processing, offering discounts of up to 90% on cached input-token reads. OpenAI also provided a 'banked reset' for Plus, Pro, and Business accounts to increase overall usage flexibility. The models are already integrated into third-party services including OpenRouter, Devin, and Venice, and follow a pattern of recent cost reductions in the frontier model market, including similar moves by Anthropic.