Anthropic has begun quietly testing its upcoming frontier model, with early examples of Claude Fable 5.2 appearing on X this weekend, a few weeks after the launch of Claude Fable 5.1. Users who shared the alleged outputs said standard prompts were generating responses that seemed much more advanced than anything Claude typically produces, though Anthropic hasn't commented on the potential A/B test. Chetaslua shared output of a ship-shooter game where Fable 5.2 made the game and the video demo, and followed up with other examples, including one posted today describing the model as "pure JS thats it one shot," noting it gave every animal its own customized snack without being asked. The clips show interactive canvas scenes where animated characters track the user's mouse cursor. None of this used external image files, plugins, or third-party tools; the model produced raw JavaScript at once and drew each sprite and animation directly in the code. Another demonstration features a custom Claude mascot walking around a small virtual world and moving to where the screen is clicked. All signs point to Anthropic quietly directing certain prompts through the web interface as it rolls out changes like merging Cowork and chat into a single experience for paid subscribers. People are comparing it with OpenAI's next major flagship product; tester JAZII said the new Fable produced entirely different results using the same complex prompts compared with first outputs from GPT-6 Astra at high reasoning effort.