Kimi K3.1 Pro vs GPT-6 Astra: same Three.js scene benchmark
Same prompt for a 28-second Three.js dragon flight: Astra took 9m17s and $2.63, Kimi 12m28s and $1.10, though a temperature bug blocked Kimi's self-review.
Kimi K3.1 Pro vs GPT-6 Astra someone tested both models with the same prompt: build a 28-second Avatar-style dragon flight in Three.js, with 6 camera shots, waterfalls, a night jungle and sunrise. > Kimi took 12m 28s and cost $1.10 > Astra took 9m 17s and cost $2.63 both generated the entire scene in one HTML file with the geometry and textures created in code. Kimi used fewer steps and even wrote its own browser testing script. Astra meanwhile could review its renders and fix issues 4 times. a temperature bug prevented Kimi from reviewing its own screenshots, so the comparison isn't entirely fair. which one would you pick?