Was this newsletter forwarded to you? Sign up to get it in your inbox.
Anthropic has historically pitched Sonnet as its “just right” model: smarter than Haiku, cheaper and faster than Opus—enough intelligence for getting work done, at a price that won’t make your accounting team cry.
Sonnet 5, released on Tuesday, arrived with an even bigger promise. Anthropic says it’s more agentic, closer to Opus, and better suited to the inchoate work people hand to AI all day.
After testing it across Every, we came away with a less flattering impression. Sonnet 5 is capable. It can produce decent prose, handle structured knowledge work, and make progress on some coding tasks. But over and over, the same question came up: When would we pick Sonnet 5 over the models already in the rotation?
We get into all of it in our Vibe Check, including:
- Kieran Klaassen’s coding results, where Sonnet 5 got stuck on agentic builds that stronger models handled better
- Katie Parrott’s writing tests, where Sonnet 5 produced usable promo copy—but its editorial instincts went sideways
- Mike Taylor’s take on decks, maps, and whether the price makes sense for knowledge work
- Austin Tedesco’s side-by-side email draft test against Opus 4.8
- Plus: Our Reach Test verdict, benchmarks, and examples across coding, writing, knowledge work, and agent behavior
On its own, Sonnet 5 is a decent model. Next to GPT-5.5, Opus 4.8, and Fable 5, though, there’s always a faster, cheaper, and more capable option.
Katie Parrott is a staff writer at Every. You can read more of her work in her newsletter.
To read more essays like this, subscribe to Every, and follow us on X at @every and on LinkedIn.
Discover Every’s upcoming workshops and camps, and access recordings from past events.
For sponsorship opportunities, reach out to sponsorships@every.to.
