OpenAI GPT-5.5 Instant: New Default ChatGPT Model

Source: OpenAI Blog Date: 2026-05-05 URL: https://openai.com/index/gpt-5-5-instant/

Summary

OpenAI released GPT-5.5 Instant as the new default ChatGPT model, replacing GPT-5.3 Instant. Three quantifiable improvements: 52.5% fewer hallucinated claims on high-stakes prompts (law, medicine, finance), 30.2% fewer words per response, and 29.2% fewer lines. Introduces personalization via search across past conversations and Gmail for Plus/Pro users. Rolling out to all users, with GPT-5.3 Instant available for three months before retirement.

Implications

• Token economics thread: the conciseness optimization (30% fewer words) directly reduces cost-per-interaction, addressing the data center margin problem (16.7% at 100% tenancy). Fewer tokens per response = more conversations per GPU-hour.
• Sycophancy convergence: both OpenAI (model-level retraining for conciseness) and Anthropic (configurable effort levels) are solving the same verbosity/sycophancy complaint from different angles. The problem is confirmed real by both vendors’ independent investments.
• Personalization via conversation history search: blurs the line between assistant and personal knowledge graph. Combined with Gemini’s chat history import tools, the stickiest product is the one that knows you best. Context portability thread.