@dabit3 Building a platform to benchmark top models (Claude, GPT, DeepSeek, Kimi, etc.) on real coding performance and actual cost per task, with Devin Max running the pipeline and automated benchmarks.
Goal: pick your AI stack with hard numbers, not hype.
@dabit3 I’m building an AI-native workflow for indie game development — localization, debugging, refactoring, testing or shipping features from a single spec. Would love to pub GPT-6 Astra + Devin on some genuinely long-horizon coding tasks