Models & Releases

ChatGPT 5.2 Thinking routed to Instant and give low-quality answer

A simple car wash question reveals a major flaw in ChatGPT's new thinking mode.

Deep Dive

Users report ChatGPT 5.2's new 'Instant' answer mode is giving low-quality, incorrect responses. When asked if you should walk or drive 50 meters to a car wash, Instant mode incorrectly says 'walk.' Only when forced to 'think longer' does it correctly state you must drive the car to the wash bay. This suggests a routing bug where quick answers bypass the model's reasoning, leading to logical failures on simple tasks.

Why It Matters

If ChatGPT can't reliably reason about basic real-world logic, its utility for complex problem-solving is in serious doubt.