Jesse Cunningham Reviews Claude Opus 5

0:00Anthropic just released Claude Opus 5. Now, it's looking pretty darn good on benchmarks against Fable 5, the previous model of Opus and GBT 5.6 Soul. It beats everything except on a few occasions. Now, this is the press release more or less from Anthropic. It just came out. The goal of this video is to break this down into simple terms to figure out is it hype or is this a big deal? Now, this is what gets me most excited. We have a re-pricing structure because Opus 5 is extremely powerful as powerful almost as Fable 5 in some regards more powerful, but it's about half the darn price. So, Claude just cut the price of intelligence in half. I like it. Excited. That's the new default on Claude Max. It's the strongest model on Claude Pro and it's live for anyone who's using Claude Code right now. Now, to demonstrate how powerful it was, they have a wind tunnel demonstration of Opus 5 building this out itself on different figures. We have a car here and then we have a figure here of the Claude icon, right?

1:00We have different things, a cell artifact, but if we come down to some quotes, I like this. So, the CEO of Cognition said on Frontier Code 1.1, Claude Opus 5 approaches Fable level performance at half the cost. And then Cursor, the co-founder says Claude Opus 5 delivers near Fable 5 intelligence at Opus speed and cost. Now, for people like us, does this actually matter? So, it's 0.5% behind the flagship at Code and this is huge.

1:27So, Opus 5 lands within a half of a percent of Fable 5. At half the cost, I'll take that little less ability in Code and for half the cost all day. Absolutely. Two times plus its own predecessor doubled, right? That's just saying it's smarter than the previous Opus model. We get that. A third of the cost for computer use. I'm excited about this one. So, on the computer use benchmark, it beats Fable 5 while costing a third as much. A third as much. That is crazy. Agents that browse click in research just got cheap. That matters for people like me who build websites and emulate other websites and aspects of CSS and blah, blah, blah. It matters a lot. 3x next best abstract reasoning. So, triple the best model score on ARK's hardest reasoning test. Basically, the figure it out muscle is three times more powerful on this model. Now, another quote here from JetBrains head of AI says it's the clearest jump in problem solving we've seen from one Claude model to the next.

2:27Powerful. It checks its own work now. It verifies before it ships. Much like Fable. I love how Fable did that. It iterates until success. Let's keep it rolling. What does this mean for the stuff we build on this channel? Microsites get cheaper. The wiki librarian costs less. Recon and agents unlocked in the pro tier now grew up. Right? Very cool. Cheaper. Like it. What does this not change? Google's not going to rank you just having a fancier model. Benchmarks are not your niche. Nothing you build needs rebuilding. Your stuff that you build is probably just fine.

2:57So, three moves all free. Check your Claude code model right now. Rerun something you ran last month with the same prompt just with Opus 5. Requeue the two expensive ideas because if you were like me using Fable, some of this stuff was hitting my limits and it was just too expensive. And I'm on the $200 a month plan. So, the tools price has changed and then the playbooks we talk about on this channel have definitely compounded. Now, for me, I was excited when Fable came out. Right? Fable came out blew my mind. Now, if you're telling me that this model can do what Fable does,

3:27pretty much at a third or half the price, I'm all in. I think this is a big announcement.