AI Release 🔥 A dispute on Habr gathered three hundred comments — and not a single benchmark. I took four models, from cheap to top-tier,

AI Release 🔥 A dispute on Habr gathered three hundred comments — and not a single benchmark. I took four models, from cheap to top-tier, and ran them through five identical routine tasks. The difference turned out not where expected: on simple scripts and edits, cheap ones responded almost like flagships, but on ambiguous formulations they started to fall apart. This is not a reason to chase the most expensive API — first check your typical