I compared GPT-5.1 to GPT-5 on ChatGPT, and now I don’t want to go back (www.techradar.com)

OpenAI quietly made GPT-5.1 the default ChatGPT model; rather than promising a dramatic leap, the release tightens and refines GPT-5’s weaknesses — and in hands-on comparisons it consistently outperformed its predecessor across instruction-following, tone, reasoning transparency, and image handling. In tests the author gave constrained tasks (a four-sentence, kid-friendly summary of The Lion King with sentence-start rules), conversational explanations of motion sickness, a practical fuel-cost math problem, and image edits/classification. GPT-5.1 nailed the syntactic constraints, delivered a warmer, more natural voice, showed clearer “work” and real-world rounding in arithmetic, and produced image edits that better preserved the subject’s face while giving crisper visual reasoning. For the AI/ML community this matters because it signals a shift from raw capability boosts toward calibration: improved alignment with user intent, more reliable constraint-following, clearer explanations, and stronger multimodal consistency. These are incremental but high-impact gains for real-world apps (UX, instruction-following agents, explainability, and trustworthy image editing) and suggest OpenAI is preparing the ecosystem and expectations for a larger architectural step in a future GPT-6.