OpenAI 取消了使用上限
openai 移除了免費用戶的文字聊天限制並推出 gpt-5.6 luna，將事實錯誤率降低了 62%，這加劇了 ai 基礎設施的軍備競賽。
openai is rolling out unlimited text chats for free users and introducing gpt-5.6 luna, which reduces factual errors by 62% compared to the previous model. paid users access the faster gpt-5.6 sol model and a 'think' slider to tune reasoning complexity, signaling a shift toward user-controlled model latency. limits remain for images and voice, but removing text barriers pressures rivals like google and anthropic to offer similar value. the update highlights a race where scale and accuracy must coexist with sustainable economics.