ChatGPT 5.5 scored 87 where the next best model scored 67. Here’s what that gap looks like in real work.

Source: Nate’s Newsletter Date: 2026-04-28 URL: https://natesnewsletter.substack.com/p/chatgpt-55-scored-87-where-the-next

Summary

A practitioner review of GPT-5.5 (the benchmark and its name appear to be the author’s framing, not an official OpenAI release designation) describes a 20-point gap over the next-best model on an unspecified evaluation, but grounds the claim in three real-work tests: executive documentation, data migration, and visualization. The author’s conclusion is that GPT-5.5 meaningfully outperforms competitors on complex multi-step professional tasks, not just synthetic benchmarks.

Implications

• Model capability / frontier benchmark thread: The 20-point gap claim, if the evaluation is credible, would represent a significant capability discontinuity — the kind that shifts which model practitioners default to for production work. Worth tracking for corroboration from independent evaluations.
• Practitioner adoption thread: The framing (“most blown-away I’ve felt”) reflects a recurring pattern where practitioner perception of a capability jump precedes structured benchmark confirmation. These early practitioner signals are often leading indicators of model adoption shifts.