OpenAI’s recent banked reset for Codex usage limits on September 22, 2026, followed the launch of GPT-6 Sol and Luna models, which delivered broad capability gains and a 50% API price cut to encourage heavier adoption. Tibo Sottiaux’s announcements tied these resets to paid tiers, compensating for rollout timing and quality fixes after earlier Astra updates. Frequent resets—averaging every 7 days recently—reflect OpenAI’s strategy to balance compute demands, smooth load via 5-hour windows, and retain users amid competition from other large language models. Traders watch Tibo’s X posts for signals on further adjustments ahead of potential policy tweaks or new model iterations.