Since August 6, 2026, ChatGPT has offered a reasoning slider on its default paid model, GPT-5.6 Sol. In practice, a user can now choose, task by task, how much the model should "think" before answering. For a small business paying for subscriptions and usage-based AI, this setting is not a minor interface detail: it is a new lever to balance speed, quality and cost.
In brief
- On August 6, 2026, OpenAI announced a ChatGPT update: a manual five-level reasoning slider for GPT-5.6 Sol, rolled out on web, mobile and desktop apps for Plus and Pro subscribers.
- This slider was already available in ChatGPT Work (the business plan) before reaching the consumer subscription.
- GPT-5.6 Luna becomes the default model for Free and Go accounts, with unlimited text chats announced for the following week, plus a more limited "Think" button compared with paid subscribers' slider.
- According to OpenAI's internal evaluations, GPT-5.6 Sol produces 68% fewer factual errors and GPT-5.6 Luna 62% fewer, compared with the older fast model GPT-5.5 Instant. These are vendor-reported figures, to be treated with the usual caution toward self-reported benchmarks.
- For a small business, the practical benefit is twofold: better control over response time and cost for each use case, and a clearer basis for choosing the right subscription for each person on the team.
What changed in ChatGPT since August 6, 2026
OpenAI publicly launched the GPT-5.6 family (Sol, Terra, Luna) on July 9, 2026, with three model tiers for three different needs. A month later, the August 6 update does not change that architecture: it adds a fine-grained control inside the Sol model, the most capable of the three.
Until now, a Plus or Pro subscriber could not choose how long the model spent "thinking" before answering: ChatGPT handled that automatically based on the perceived complexity of the question. With this update, the user takes control. The new slider lets you manually set the reasoning effort, from the fastest setting (near-instant answer for a simple question) to the deepest (thorough analysis for a complex subject: code, planning, strategic decisions).
This feature already existed in ChatGPT Work, the business-focused plan, before being extended to the consumer subscription. It only applies to ChatGPT: GPT-5.6 Sol remains unchanged in Codex, OpenAI's dedicated coding environment.
The reasoning slider, in practice
Open GPT-5.6 Sol
Choose the reasoning level
Raise the slider for high-stakes tasks
Adjust based on the result
Before August 6, 2026
Since August 6, 2026
What this means for a small business
The slider does not change the price of the ChatGPT subscription itself: Plus and Pro stay at the same rates. But it gives a team a direct way to manage response time and usage of the most capable model in the lineup. A small business using ChatGPT for everyday writing, meeting prep or document analysis can now:
- reserve the deepest setting for tasks that genuinely need it (contracts, cost estimates, strategic plans);
- keep the low setting for everyday exchanges, which are faster to get;
- train the team on this choice instead of relying only on an invisible automatic setting.
This logic echoes a principle already covered in our article on AI cost governance: the clearer a leader's guidance on when to use a model's maximum power, the less an AI bill drifts by default of judgment.
Accuracy claims: figures to take with caution
OpenAI highlights a significant improvement in factual reliability compared with the older fast model, GPT-5.5 Instant.
Reduction in factual error rate as reported by OpenAI, compared with GPT-5.5 Instant (vendor's internal evaluation, not independently verified).
A limit worth knowing for free accounts
The "Think" button offered to Free and Go accounts is not equivalent to paid subscribers' slider: it extends GPT-5.6 Luna's thinking time but does not switch to a more capable model. For high-stakes topics (legal, financial), it is worth double-checking the answer or moving to a paid subscription.
ChatGPT plans compared after the update
| Plan | Default model | Reasoning slider | Text chats |
|---|---|---|---|
| Free / Go | GPT-5.6 Luna | "Think" button (extends thinking time, no model switch) | Unlimited per the announced rollout |
| Plus / Pro | GPT-5.6 Sol | Manual 5-level slider | Unlimited per subscription |
| ChatGPT Work (business) | GPT-5.6 Sol | Already available before the consumer rollout | Unlimited per license |
FAQ
What is the GPT-5.6 Sol reasoning slider?
It is a manual five-level setting, added on August 6, 2026 in ChatGPT, that lets Plus and Pro subscribers choose how much time and effort GPT-5.6 Sol spends on an answer, from a fast response to a thorough analysis.
Does the "Think" button for free accounts do the same thing as the slider?
No. The "Think" button available to Free and Go accounts extends GPT-5.6 Luna's thinking time, but does not grant access to a more capable model like GPT-5.6 Sol.
Does this update change ChatGPT subscription prices?
No, Plus, Pro and Work subscription prices remain unchanged. The update concerns the settings available within these subscriptions, not their price.
How should a small business use this setting day to day?
By reserving a high reasoning level for high-stakes tasks (contracts, financial analysis, strategic decisions) and a low level for everyday exchanges, to get faster answers without sacrificing quality on the topics that matter.
Conclusion
GPT-5.6 Sol's reasoning slider illustrates a broader trend among generative AI vendors: giving users explicit control over the speed-quality-cost tradeoff, rather than deciding it for them. For a small business, this is more than an interface gadget: it is a concrete tool to frame how the team uses ChatGPT, without changing subscriptions. It remains to be seen, over the coming weeks, whether OpenAI's announced accuracy gains hold up in real-world use.
To go further, browse our other AI resources for leaders and discover real company case studies structuring their day-to-day AI use.
