gpt-6IMAGE
Cat-counting test: GPT-6 Astra gets 17 cats right, GPT Image 2.5 fails
In a cat-counting comparison run in ChatGPT Instant mode, GPT-6 Astra answered 17 cats correctly while GPT Image 2.5 got it wrong.
Elaina
@ElainaNotesArchived 2026.09.14Originally posted 2026.09.09

Elaina @ElainaNotesOriginally posted 2026.09.09
GPT Image 2.5 failed this cat counting test. GPT-6 Astra got it right: 17 cats. 😹🐈 I specifically used ChatGPT in Instant mode to avoid any extra Thinking from another model interfering with the test. So for this round of cat counting: GPT-6 Astra ✅ GPT Image 2.5 ❌
This post quotes another post: “GPT-6 Astra passes the cat test”
x.com/elainanotes/status/2097439453707206873Archived 2026.09.14 16:20