OpenAI restricts free users to Luna as Sol gets upgraded
OpenAI's model tiering concentrates its best reasoning among paid professional users, raising the bar for what content earns citations.
Key takeaways
- Free ChatGPT users are restricted to GPT-5.6 Luna from next week; paid users get the upgraded Sol with a reasoning slider.
- Sol's reasoning slider makes the model more deliberate and source-critical at higher settings, raising the bar for cited content.
- The paid tier skews toward professional and institutional users — the B2B audience that most brands need to reach via AI answers.
- Content that is thin or gated will underperform against Sol's deeper evaluation; structured, authoritative public content benefits.
- Brands cannot control the reasoning depth a user sets, but they can control whether their published record is worth citing at maximum depth.
Free users of ChatGPT will soon interact exclusively with GPT-5.6 Luna, OpenAI's smallest model. The Decoder reports that OpenAI is restricting free-tier access to Luna starting next week, while simultaneously upgrading GPT-5.6 Sol with more focused responses and a reasoning slider that lets paying users calibrate how deeply the model thinks before it answers. The gap between tiers is not cosmetic.
This is a tiering decision with direct consequences for how brands appear in AI-generated answers, and who sees those answers.
The mechanics of the split
Sol now sits above Luna by a meaningful margin. The reasoning slider is the telling detail: it gives Plus and Pro subscribers a way to trade latency for depth, pulling Sol toward the kind of deliberate, evidence-heavy responses that tend to cite structured, authoritative sources. A model prompted to reason longer does not reach for the first plausible answer; it evaluates. That evaluation process is where source credibility operates as a filter.
Luna, meanwhile, gets unlimited text chats and a button that lets it reason longer. OpenAI is careful to present this as a feature addition for free users. The substance is different. Luna still falls well short of Sol, per The Decoder's account, and a "reason longer" button on a weaker model is not the same as a more capable model reasoning by default.
Who is actually asking the questions that matter to B2B brands
Free-tier ChatGPT users are not the buyers that financial institutions, UN agencies, or industrial procurement teams send to negotiate contracts. The professionals running policy research at a multilateral, scoping a supplier for a major infrastructure project, or evaluating a fintech vendor are, almost universally, operating on paid accounts. They get Sol, the slider, and deeper reasoning chains.
This matters for content strategy in a specific way. A model instructed to reason more carefully before answering is more likely to surface sources it can verify and attribute. It is less likely to hallucinate a citation and more likely to pull from documents, reports, and structured web content that signals expertise. Brands whose content is thin, vague, or buried behind registration walls will not fare well under that scrutiny. Brands with a published record of substantive analysis will.
The asymmetry runs in both directions. If a brand's best content is authoritative and well-structured, Sol's improved reasoning is an opportunity: the model is more likely to find it and cite it accurately. If a brand's public content is sparse, the same reasoning process is more likely to surface a competitor who has done the work.
The signal for content investment
The reasoning slider also raises a practical question about query intent. When a senior procurement officer at a large industrial group asks ChatGPT to compare suppliers in a given category, the depth of Sol's response will depend partly on how much the slider is set. At higher settings, Sol will draw on more context and weight source quality more heavily. At lower settings, the answer will be faster and shallower. Brands cannot control the slider, but they can control whether their published material gives the model something worth citing when the slider is up.
The structural implication is straightforward. OpenAI is concentrating its most capable reasoning in the hands of users who pay, and those users skew toward the professional and institutional. That is exactly the demographic that B2B brands need to reach through AI-generated answers. Investing in content that performs under deliberate model reasoning is not a hedge against future changes; it is a response to a change that has already been made.