Why is Grok Bot reaching for Claude Opus 5.5?

A rival's model inside your own product is a strange move. Unless there is more going on.

Two weeks ago, the Grok camp was pitching its latest model as the strongest one around for coding. Now the message is: Grok Bot will pick the best model for each job. And the names on the list include Claude Opus 5.5, plus Midjourney for images and Suno for music. A member of the team added that the bots will be powered by Opus 5.5.

Slow down though. No start date, no pricing, no map of which tasks go to which model. And Anthropic, Midjourney and Suno have not publicly confirmed anything. So this is an announcement, not a rollout.

A loop

Here is the twist. Since May, Anthropic has been using Colossus, the huge GPU cluster on the Grok side: more than 300 megawatts and over 220,000 GPUs. So one side rents the muscle, and now the other side wants to borrow the brain. Round and round.

Why not just use your own model? Is this pride losing to practicality?

There are two ways to read it. The practical one: routing each task to the best model is where the whole market is heading, because nobody is best at everything and users only care about the result. The less flattering one: for some jobs, they simply trust Claude more.

The honest answer is that we do not know yet. What we can see is that the old picture of rival camps is getting blurry, and computing power is turning into a shared resource.