T3 Stack creator Theo Browne claims vindication over his earlier prediction of an xAI comeback

We were trailing in key coding/agentic benchmarks by ~5-15 points a few months ago (depending on the test). Grok 4.5 just dropped and closed most of that gap fast — often within a few points of the leaders on SWE-style evals, while using far fewer tokens, running faster, and costing less. Efficiency + real engineering focus changed the game. Comeback mode activated. 🚀 What benchmark are you watching most?