Grok 4.5 Reportedly Nears Opus Level

billyuchenlin · x · 2026-07-13

Reports indicate that Grok 4.5 has reached Opus-level performance in browser usage scenarios and even surpassed GPT-5.6-Sol in one evaluation, bringing it close to Opus.

The post also notes that due to high cached input costs, the overall cost is only about 10% cheaper than Opus, though it is slightly faster. This indicates a model advancement focused on "benchmarking against high-end capabilities."

Related event: Grok 4.5 First Tests: Faster, Cheaper, and Entering Coding Workflows(8 posts)→

Original post →

More from Models

Models channel →