I wanted to see how far Grok 4.6 would take a simple creative prompt.
Same instruction. Opus 5, GPT-5.6 Sol, and Qwen 3.8 Max joined in. Grok turned the banana into a tiny island. The different interpretations were way more interesting than I expected.
Grok 4.6 beats Claude Fable 5, Claude Opus 4.8, Gemini 3.1 Pro, GPT-5.5 and GPT-5.6 Sol on EEBench, an electrical-engineering benchmark.
Made another game with Claude's Opus 5, using the Artifacts feature 🤗
This one is called Monster Maze: The Nutrient Quest. You work your way through a maze toward the goal, curing the ailments of imaginary monsters 👻 with capsules, supplements, and fruit juices 🥤 along the
[I tried AI-driven development] AI development environment evolution observation diary #5: I finally got to the video
In Gen0, the development environment had only 3 files: README.md, AGENTS.md, and CLAUDE.md, but after 5 development requests, Markdown/note articles →PPTX → Slide image → Narration → MP4 video It was connected to
You have reached the end of the archive
All of Claude Opus 5