Matt Wolfe Tests Opus 5.5 for 20 Hours: Possibly the Best Coding Model Now

Matt Wolfe · youtube · 2026-09-30

YouTuber Matt Wolfe ran Claude Opus 5.5 on his Megabonk game test for nearly 20 hours and calls the result kind of insane — possibly the best coding model available right now. It's included in Claude's Pro, Max, Team, and Enterprise plans starting at $20/month Pro; API pricing is $4 per million input tokens and $20 per million output. He finds that price-to-performance wild for long-running coding tasks.

Original post →

More from Models

Models channel →