Lates News

OpenAI Codex Product Lead Tibo: Day-one update we've optimized the default speed of GPT-6 Astra and GPT-6.1 Sol, for an overall speedup of about 50%. This optimization covers all subscription users, as well as all products and partners that sign in with ChatGPT accounts. No action is required on your part, and the optimization is expected to take effect within two hours. Model throughput has increased from 30 TPS to 50 TPS, paired with the best tokenizer currently available. And in terms of the number of tokens required to complete a task, these models are remarkably efficient.