What GPT-5.2 means for software engineers
GPT 5.2 just dropped, and it crushes every software engineering benchmark
Gemini is no longer king. Opus 4.5 isn’t even on top anymore.
GPT-5.2 dropped today and the benchmarks are off the charts. SWE-bench pro (coding): comes in at 55.6% compared to Gemini 3 Pro’s (43.3%).
But you’ve probably seen the benchmarks, and you’re probably wondering why the model matters to you. In today’s newsletter, we’ll get into the details of why GPT-5.2 matters for:
One-shotting whole apps
Code generation
Reviews and refactors
Autonomous agents
Massive context windows
Images, spreadsheets, and data
GPT 5.2 is really good at one-shotting simple apps
This was a major theme of press release.
Here’s a prompt that they shared for a holiday card maker app, followed by what the model came up with:
Create a single-page app, in a single HTML file, that demonstrates a warm and fun holiday card! The card should be interactive and enjoyable for kids!
- Have variety of items kids can drop in the UI; a few should be already placed by default
- Also have fun sound interactions
- Place many cute and fun stuff as much as possible
- Animation like snowdrop should be used nicely
Here’s another set for a typing game:
Create a single-page app in a single HTML file with the following requirements:
- Name: Typing Rain
- Goal: Type falling words before they reach the bottom.
- Features: Increasing difficulty, accuracy tracker, score.
- The UI should be the city background with animated raindrop words.
This is really cool, and reinforces that we’re entering the age of personal software, where you can build any low-medium complexity app incredibly fast, regardless of how small the commercial market is.
But if you’re working in software professionally, you probably care more about how the model performs for your day to day work.