June 12, 2026
1 min read
A few days ago Anthropic released the model that was supposedly too dangerous for the world to see. I wanted to give the model a try and used my personal benchmark: Can it create an idea for a game that I've had for years in one shot. The model started with a -- very -- long reasoning session and after 45 minutes and more than 20 euros worth of tokens (!) it happily reported that "Shepherd's Dog is complete — a single 1,296-line index.html with zero dependencies."
You can see the result in the video below or give the game a try It's really fun and exactly how I imagined it. For me this is the first time an AI model is able to create this for me in one go:
- Other attempts with earlier models:
[github.com/vnglst/when-ai-fails/blob/main/shepards-dog/README.md](https://github.com/vnglst/when-ai-fails/blob/main/shepards-dog/README.md)
source & further reading
koenvangilst.nl — original article
Notes from the Mistral AI Now Summit in Paris