Why has auto mode disappeared?

Something has been quietly disappearing from ChatGPT's interface since late March, and the pattern of reports is now too consistent to dismiss as individual glitches. The first signal arrived on March 22nd: a user asking why "auto mode" had vanished — the setting that let the model judge for itself whether a prompt warranted deep reasoning or a quick answer. That feature carried a quiet design intelligence. It meant the system could exercise its own epistemic judgment, deciding how hard to think without being told. Small to look at. Significant in what it said about how OpenAI wanted the product to behave. Then, on April 30th, a second thread surfaced asking about a "fast mode" toggle that had apparently also been promised and then removed. Two separate features, two separate user expectations, both quietly gone. By May 4th, something more noticeable was being reported: answers arriving completely instantly, with no thinking time displayed at all — the kind of speed that suggests not a UI tweak but a behavioural change underneath. Then on May 7th, OpenAI's own platform notice confirmed what users had been circling: both the "Thinking" mode and "Memory" features had disappeared from the ChatGPT iOS app. That last signal shifts this from speculation to something with an official footprint, even if the reasoning behind the changes remains unexplained.

If confirmed as a deliberate strategic direction rather than a phased rollout glitch, this matters more than a settings menu change would suggest. Auto mode and thinking mode together represented a particular bet on user experience — that users would trust the model to self-regulate its own reasoning depth rather than needing manual control. Removing those features, and replacing apparent deliberation with instant responses, signals either a performance optimisation play or a product simplification strategy aimed at a broader, less technically curious audience. The second-order effect is real: power users who relied on visible reasoning to calibrate their trust in outputs now have less signal about what the model is actually doing under the hood. Memory disappearing from iOS compounds this — not just a feature loss, but a continuity loss, making the app feel more transactional and less relational. If OpenAI is consolidating its interface to reduce complexity before a major capability release, this pattern would make sense. If they are simply deprecating features that underperformed in usage metrics, that is a different story about what people actually want from the product.

Watch for any official OpenAI changelog or product blog entry explaining the removal of these features — and pay close attention to whether Memory returns in a redesigned form, which would suggest a rebuild rather than an abandonment.

NewsHive monitors these sources continuously. All signal titles above link to the original reporting.

Intelligence by NewsHive. Need help navigating what this means for your business? Contact GeekyBee →