Same prompt, different results: Claude Opus 5 vs ChatGPT Astra image generation
The author gave Claude Opus 5 and ChatGPT Astra the same prompt to create a kawaii background for a product photo and says the gap between the two outputs is obvious.
Imagine thinking Claude could even compete with ChatGPT in any sense of the term. I used the same exact prompt for these two images, with as equally comparable models as possible. Claude Opus 5 (high) on the left vs ChatGPT Sol 5.6 (high) on the right I told them to recreate a kawaii background for this product photo. That's it. That's all. Speechless. Now, cue all the Claude users in the comments saying: "bUt ClAuDe iS bEtTeR aT OtHeR ThInGs iT jUsT sUcKs At ImAge GeNeRaTiOn" No, it absolutely isn’t. Claude has been dramatically worse at everything I’ve EVER used it for, but the gap becomes genuinely embarrassing when you compare Fable with Astra on complex work now. It's not even in the same realm. Even more importantly, why the heck would you even RISK using Claude for anything else after seeing this output in one area of it's capabilities? When people talk about how awesome Claude is and/or that they use Claude primarily, I immediately know they haven't explored ChatGPT Astra (or even prior models) thoroughly yet. When you say “Claude is the best,” what you actually mean is: “Claude is the one I know how to use, and I haven’t left my comfort zone long enough to seriously or unbiasedly test anything else.”