Claude Opus 5 first impressions
Anthropic's new Claude Opus 5 model is here, and I don't think I need to use the Fable 5 model anymore for basically anything. The new model is already better than Fable 5 in almost every aspect, as you might see in the benchmark below.
And the best thing? It's priced at the half of the Fable 5 model, the same as how previous Opus models like 4.8 where priced. So it does make sense when I say that there's no need to use Fable 5 anymore, no?
I have been using the model for the last ~30 minutes inside Claude Code, and so far I like it. I am keeping it at the Max effort and the limits are still not draining too fast like Fable. And like other Opus models, the new one is also great at frontend designs.
I will share more info as I use and experience the new model.
Update: July 25, 2026
One thing I noticed is, Opus 5 is consuming significantly fewer tokens than the previous Opus 4.8 model for the same tasks. This is confirmed in their launch post as well that Opus 5 has a lower cost per task than 4.8.
My $100 Claude plan is very usable once again.