Perplexity swaps in Grok 4.5 for its Computer feature
Cheaper model takes over agent coordination on Pro and Max plans
@perplexity_ai3 days ago
TL;DR:
- Grok 4.5 now runs Perplexity's Computer product for paid users
- Company claims it beats other setups on internal tests and costs half as much
- Still just their own numbers until someone else checks
Perplexity's Update
Perplexity is using Grok 4.5 to run its Computer feature for users on the higher plans. They claim it beats other setups on their WANDR test and costs about half what Opus 4.8 does.
The idea is that these orchestrator models handle the planning and tool calls for AI agents. Perplexity thinks Grok 4.5 is better and cheaper than the Anthropic option. That could matter for anyone running agents where the bill for running the model adds up fast. Still, it's their own test, so take it with a grain of salt until others check it out.
What It Means
Medium significance. It's a product update in the AI agent space.