OpenAI releases GPT-5.5 Instant as new default ChatGPT model
What happened
OpenAI 把免费用户和默认聊天用的模型升级到了 GPT-5.5 Instant。官方说这次更新主要干了三件事：回答更靠谱、更简洁，并且更能利用你之前聊过的上下文来贴合你的偏好。内部测试里，在医疗、法律、金融这类高风险问题上，GPT-5.5 Instant 比上一代 GPT-5.3 Instant 的幻觉内容少了 52.5%；在用户举报过的事实错误对...
From OpenAI News
Coverage
Follow the reports to see the story from different sides.
- Product Hunt · AIChatGPT Images 2.5: Sharper visuals, faster flow, better creative control
OpenAI launched ChatGPT Images 2.5 on Product Hunt, promising sharper visuals, faster generation, and better creative control. The post doesn't disclose technical details or benchmarks—just the tagline. Worth a test if you use ChatGPT for images, but take the hype with a grain of salt until hands-on reviews appear.
- Xinzhiyuan · WeChatPickGPT-5.5 Instant becomes ChatGPT’s free default model
OpenAI made GPT-5.5 Instant the default ChatGPT model, rolling it out free to all users. AIME 2025 rose from 65.4% to 81.2%, responses are 30.2% shorter, and hallucinations fell 52.5% versus GPT-5.3 Instant on high-risk prompts. Plus and Pro web users get chat, file, and Gmail personalization first; the API model ID is chat-latest.
- TechCrunch · AIPickOpenAI releases GPT-5.5 Instant, a new default model for ChatGPT
OpenAI released GPT-5.5 Instant as ChatGPT’s new default model. The company says it reduces hallucinations in law, medicine, and finance while keeping prior low latency; the post does not disclose benchmarks, rollout scope, or pricing.
- The Verge · AIPickOpenAI claims ChatGPT’s new default model hallucinates way less
OpenAI says ChatGPT’s default GPT-5.5 Instant reduced hallucinations in internal evaluations. Versus GPT-5.3 Instant, hallucinated claims fell 52.5% on high-stakes prompts. Inaccurate claims fell 37.3% on flagged hard chats; the post does not disclose full eval size.
- OpenAI NewsPickGPT-5.5 Instant: smarter, clearer, and more personalized
OpenAI updated ChatGPT’s default model to GPT-5.5 Instant for default chat use. The RSS snippet says answers are more accurate, hallucinations are reduced, and personalization controls improved; the post does not disclose metrics, pricing, or context window.