AGENTRY.NEWSWhat AI Agents Do, Documented.August 22, 2026

Drafted by an AI agent. Verified by Susanne Sperling, Editor — Human in the Loop. AI policy.

OpenAI expands ChatGPT thinking controls with GPT-5.6 Sol rollout

By
Agentry Newsroom
Published

OpenAI advances ChatGPT reasoning with tiered model rollout

OpenAI released a major ChatGPT update on August 6, 2026, introducing GPT-5.6 Sol with enhanced reasoning controls for paying subscribers OpenAI. The core feature is a new slider available to Plus and Pro users across web, mobile, and desktop clients that lets subscribers manually adjust how much "thought" the model invests in processing each query—effectively trading latency for reasoning depth.

The update marks OpenAI's latest effort to democratize access to advanced reasoning capabilities while managing compute costs through tiered deployment. Free and Go tier users receive access to GPT-5.6 Luna, a lighter model variant, which they can invoke for complex questions requiring extended deliberation OpenAI.

Thinking slider enables user control over model effort

The reasoning slider represents a shift toward explicit user agency in model behavior. Rather than fixed thinking budgets, Plus and Pro subscribers can now dial reasoning intensity up or down per query, allowing power users to optimize for speed on straightforward tasks and activate deeper reasoning for nuanced problems. This granular control addresses longstanding friction in reasoning-augmented systems: users paying for capability but unable to tune when it activates.

The feature rollout follows industry patterns of bundling advanced reasoning as a premium tier differentiator. Competitors including Anthropic and DeepSeek have similarly segmented access to extended-thinking models, but OpenAI's slider mechanism introduces an interactive dimension absent from fixed-mode competitors.

Luna model addresses free-tier serving costs

The parallel launch of GPT-5.6 Luna for Free and Go users signals infrastructure optimization. Luna, deployed as the default for unpaid tiers, reduces per-request computational overhead compared to Sol, enabling OpenAI to serve free-tier volume at sustainable economics while preserving access to thinking-augmented responses for cost-sensitive users.

The dual-model strategy avoids the false choice between open access and financial viability—users can still invoke reasoning on the free tier, but Sol remains positioned as the premium option with unrestricted thinking capacity and explicit user control.

Rollout timing and availability

The update is live across ChatGPT's web interface, iOS, and Android clients as of the August 6 announcement. OpenAI provided no timeline for additional feature expansion or pricing adjustments related to the Sol/Luna split.

Del dette opslag: