OpenAI unlocks unlimited chats for free users and adds a reasoning slider

Hitting a rate limit in the middle of a productive chat is one of the quickest ways to ruin a workflow.

With ChatGPT recently crossing the 1 billion weekly active user mark, OpenAI is rolling out a massive shift. Free users are getting unlimited text conversations, while paid subscribers are gaining granular control over how long their model thinks before it speaks.

A thinking slider and way fewer hallucinations

For Plus and Pro subscribers, OpenAI refreshed the GPT-5.6 Sol model inside the standard Chat interface. The headline addition across desktop, mobile, and web is a new slider that lets you dial in the exact amount of reasoning you want per prompt.

If you just need a quick email polish or a factual lookup, keep the slider low for an instant response. If you are outlining multi-step research, debugging code, or evaluating business plans, slide it up to grant the model extra processing time. Best of all, OpenAI unified the Instant and Thinking behaviors, meaning the model keeps a consistent voice rather than feeling like an entirely different persona when it reasons.

OpenAI also tuned Sol to be far more concise, stripping out superfluous formatting and pushing back with corrections rather than simply nodding along. In internal benchmarks covering dense financial, legal, and medical queries, responses with at least one factual error dropped 68% compared to GPT-5.5 Instant. (The lighter GPT-5.6 Luna saw a 62% drop.)

Keep in mind that this Sol update only touches general Chat. The models running under the hood for Codex and Work remain untouched. OpenAI also rolled out stricter system-level guardrails for users identified as under 18, cutting off romantic roleplay and enforcing stricter safety boundaries around body image, dangerous stunts, and sensitive goods.

Free text goes infinite, with a catch

For Free and Go tier accounts, the default engine is shifting this week to GPT-5.6 Luna. Starting next week, OpenAI is dropping the rate limit on text chats entirely.

That means you can type back and forth indefinitely without getting locked out. For tricky logic or dense questions, free users will also see a new Think button next week, giving Luna permission to chew on the prompt a bit longer before replying.

There is, of course, a boundary: file attachments, vision uploads, voice mode, and image generation will still carry hard usage caps. OpenAI is generous with text tokens, but multi-modal compute remains gated.

Why it matters

We think this update signals that standard text generation has officially become cheap enough to treat as running water.

For everyday users and students, an infinite text window paired with an on-demand Think button makes ChatGPT an unbeatable default research scratchpad. For power users paying twenty dollars or more a month, the slider solves an annoying friction point: you no longer have to guess whether the chatbot is going to overthink a simple question or rush through a complex logic puzzle.

Across the AI landscape, this puts intense pressure on rivals. By removing text friction for its 1 billion weekly visitors, OpenAI raises the floor for what a free tier looks like. Any competitor still charging for basic text querying is going to feel an immediate squeeze.

At the same time, OpenAI’s strategy is clear. Pure text is the ultimate retention hook, but high-cost workloads like document processing and visual reasoning remain firmly behind the paywall. It is a calculated, aggressive move to dominate everyday user habits while keeping infrastructure margins protected.