OpenAI's new ChatGPT model claims to hallucinate 52.5% less

OpenAI has announced that its newest default model for ChatGPT, called GPT-5.5 Instant, is significantly better at telling the truth. The company claims this model produces 52.5% fewer made-up facts, or "hallucinations," than previous versions. Hallucinations are a common problem in AI, where models confidently share incorrect information. OpenAI says this improvement comes from "internal evaluations," though they haven't shared full details yet.

This matters because it could make AI assistants more trustworthy for everyday tasks. Imagine asking your AI for medical advice or financial tips - fewer hallucinations mean you're less likely to get bad information. While no model is perfect, this update suggests we're moving toward more reliable AI helpers.

If you use ChatGPT regularly, you might notice this improvement soon. OpenAI is rolling out GPT-5.5 Instant as the new default model. Try asking it factual questions and see if the answers feel more accurate. Keep an eye out for any persistent errors, as real-world testing will reveal how well this works in practice.