OpenAI has released its long-awaited GPT-5 model to all tiers of ChatGPT, saying the system offers stronger reasoning, faster coding assistance and automatic switching into a high-compute “thinking” mode. The rollout comes as the company reports about 700 million weekly active users on ChatGPT, underlining its lead in consumer AI services. The new model is subject to tiered rate limits—10 GPT-5 messages every five hours for free users, 80 messages every three hours for Plus subscribers and unlimited access for Team and Pro plans. An autoswitcher failure on 8 Aug made responses appear weaker and prompted a wave of complaints. Chief Executive Officer Sam Altman acknowledged a “bumpy” launch, doubled message caps for Plus users and reinstated the earlier GPT-4o model as an option for paying customers. Performance tests have intensified debate over the model’s capabilities. Independent evaluations on the ARC-AGI-2 benchmark show xAI’s Grok 4 scoring 15.9 percent versus GPT-5’s 9.9 percent. Grok 4 also defeated Google’s Gemini 2.5 Pro in the Kaggle AI Chess semifinal, highlighting rapidly shifting standings among so-called frontier models. xAI has moved to capitalise on the momentum by making Grok 4 free—with five requests every 12 hours—and its iOS application has climbed to the top of Hong Kong’s productivity charts. The competitive flare-up underscores how incremental product decisions and benchmark results are shaping user sentiment in the global race to dominate next-generation AI systems.