OpenAI improves GPT-5.6 Sol in ChatGPT and restricts free users to its weakest model

What happened

OpenAI updated the GPT-5.6 Sol model in ChatGPT, making its responses more focused and introducing a reasoning slider that lets users control how deeply the model thinks through problems. At the same time, OpenAI will limit free users to the smaller GPT-5.6 Luna model for text chats starting next week. Free users will still get unlimited access to Luna, along with a new button to extend its reasoning time. However, Luna remains much less powerful than the Sol model and other larger versions.

Why it matters

This update changes the value proposition for different ChatGPT user segments. Paying customers using GPT-5.6 Sol get sharper, more deliberate AI interactions, improving output quality for tricky or complex tasks. Meanwhile, free users face restricted access to a weaker model, effectively creating a two-tier experience. This pressure nudges free users to upgrade if they want better reasoning and broader capabilities.

The reasoning slider is a practical addition. It lets users dial up model focus when they need detailed, high-quality answers and dial it down for quicker, looser responses. Builders and operators should see this as a way to customize AI behavior without switching models entirely.

The fact that OpenAI is capping free use to a significantly weaker model (Luna) highlights a tightening in AI tool access and monetization. For startups or builders relying on free tiers, this may raise costs if they need higher-quality output or more nuanced reasoning.

What to watch next

Keep an eye on how this access stratification influences migration from free to paid plans. If Luna’s limitations frustrate users, OpenAI could see a faster conversion funnel, but also push some to seek alternative open or cheaper models. Watch for feedback on the reasoning slider’s usability and whether it becomes a standard feature across more OpenAI models.

Also watch for potential ecosystem shifts, such as third-party developers incorporating reasoning controls or layering smaller/larger models for cost-quality tradeoffs. OpenAI’s approach sets a precedent for how much control operators can expect over AI reasoning depth versus raw power.

AI Quick Briefs Editorial Desk