Anthropic just made Claude Sonnet 5 the default model for all Free and Pro users worldwide, and I’m not sure it was the right call – at least not yet. The switch happened on 1 July, according to Build Fast with AI, with Anthropic positioning Sonnet 5 as its most capable mid-tier model to date, close to flagship Opus 4.8 performance but at a lower price point, with introductory pricing locked in through August 2026. On paper, that sounds like a smart move. In practice, it’s a bit more complicated.
I use Claude every day for work (I’m a Software Engineer) so I’ve had a decent amount of time across different models to form a real opinion here.
Sonnet 5 is said to bring near-Opus quality at a fraction of the cost, making it the ideal everyday model for most users. Anthropic has also framed the launch as part of its broader push toward profitability ahead of a rumoured IPO later this year, per Build Fast with AI. Making the mid-tier model the default and charging less for it, is a way to grow the user base while keeping unit economics looking healthy.
It’s a business decision dressed up as a product upgrade. That doesn’t make it wrong, but it’s worth being clear-eyed about what’s driving it.
Here’s my honest experience so far: Sonnet 5 feels like it uses just as many tokens as Opus 4.8, maybe more. That surprised me. One of the trade-offs I’d accepted with Opus was that it’s thorough to a fault: it’ll check every edge case, double back on itself, make sure nothing breaks. That thoroughness costs tokens, and sometimes it’s overkill for a simple task. But the quality is undeniably there.
Sonnet 4.7, by comparison, was noticeably lighter. Quicker, less thorough which is fine for straightforward tasks, but you could feel the ceiling on anything complex and I had issues when it couldn’t find something during my planning phase of an Epic.
Sonnet 5 sits somewhere in between, but the token usage doesn’t seem to reflect that middle ground yet. If I’m burning through a similar number of tokens as Opus 4.8 but I may have to test some more before I can say with certainty here.
To be fair, it’s only been a couple of days. I’m going to keep testing it before drawing any firm conclusions. But my gut says it’s a bit early to make this the default for everyone especially paying Pro users who may have been happily running on Opus without realising the switch happened. For those who were running the older Sonnet, if it does use more tokens, then some people might get grumpy about it too.
The introductory pricing window closes at the end of August 2026, which is when we’ll get a clearer picture of what Sonnet 5 actually costs at scale. That’s also around the time Anthropic is expected to be sharpening its IPO narrative, so watch for whether Sonnet 5 adoption numbers get cited as a growth metric. If token usage stays high while price stays low, something has to give eventually.
For now, if you’re a Claude Pro user and Opus 4.8 was working well for you, it’s worth checking your settings to see what model you’re currently running on. You might not have noticed the switch.
