Users Report ChatGPT Slower, Sharper Responses; GPT-5.6 Rumored

Users on X reported that ChatGPT produced sharper outputs while taking far longer to respond this week, and several testers believe OpenAI is quietly A/B-testing a GPT-5.6 model for some Pro accounts. Reports include improved one-shot web design and more polished 3D game generation alongside response times ranging from minutes to more than an hour.

Developer Anshu Chimala posted a side-by-side video comparing one-shot landing pages and wrote ‘Well well well, I’m one of the lucky ones with early GPT-5.6 Pro access.’ Tester Conor Dart posted that a single-prompt browser game with physics and camera controls took ‘literally 60 minutes and 15 seconds’ to generate, compared with about 10 minutes on GPT-5.5 Pro. AI tester Chetas Lua reported longer runtimes for a robotics simulation and noted ‘it started to take 20-40 mins again like it used to do before 5.5 pro.’

Not all comparisons favored the rumored model. An AI benchmarker identified as Chris ran a spaceship-building prompt and recorded 87 minutes for the suspected GPT-5.6 variant versus 34 minutes and 42 seconds for GPT-5.5 Extra High. He described his expectation that GPT-5.6 would represent an ‘incremental/solid improvement’ rather than a decisive leap.

A post attributed to Pankaj Kumar listed technical details circulating among testers: a knowledge cutoff moved to December 2025, a raised internal reasoning parameter some call the ‘Juice Value’ from 768 to 960, and enhanced SVG and 3D design generation. Testers have used the nickname ‘Kindle-Alpha’ for a release candidate.

Several testers say ChatGPT appears to substitute the suspected model for a subset of Pro users when they select GPT-5.5 Pro, a form of A/B testing. An AI influencer known as Leo posted a thread claiming the suspected model is being stealth-tested in that way for some accounts and that a public launch may be planned for June 25.

OpenAI has not confirmed whether GPT-5.6 is in tests and did not respond to requests for comment.

Market and technical conditions noted by testers and observers include a recent export-control order that required a rival to suspend access to its most powerful models for foreign users, and a Chinese open-source model that has closed some performance gaps and in some benchmarks outperformed GPT-5.5 on long-form engineering tasks. Prediction market contracts for a late-June launch have traded at high odds this week.

An internal remark attributed to OpenAI chief scientist Jakub Pachocki reportedly told staff the next model is a meaningful improvement over GPT-5.5. Testers who posted examples described stronger multi-step reasoning and richer design generation, along with slower completion times on some complex tasks, suggesting differences in tuning or front-end handling.

Only OpenAI can confirm whether GPT-5.6 is being tested or when it might be released. For now, developers and researchers sharing screenshots and elapsed-time measurements on X provide the most detailed public record of what some Pro users are experiencing.