Same Boat-Game Prompt Puts GPT-6 Astra, DeepSeek-V4.1-Flash, Claude Fable 5 and Qwen3.8 Under Scrutiny

A public side-by-side demo gave Astra, DeepSeek 4.1, Fable 5 and Qwen3.8 the same boat-game prompt, turning a small creative-coding challenge into a practical comparison of current frontier models. The result is not a controlled benchmark, but it reflects where model competition is moving: from answer quality toward end-to-end software, game and agent workflows.