Anthropic’s Claude Sonnet 5 landed on June 30, 2026 with a lower price tag and performance that crowds its flagship Opus model. Both claims are true — and both hide the real story. Because Anthropic quietly shipped a new tokenizer (the same text now counts as more tokens) and trained the model to run roughly three times as many autonomous loops, one analysis found the cost to finish an actual task nearly doubled versus the last Sonnet — landing above Opus on the hard jobs where you’d reach for something cheaper.

This episode unpacks what Sonnet 5 can really do, why “cheaper per token” has stopped meaning “cheaper per job,” how its agentic training makes capability and cost the same design choice, and why Google’s Gemini 3.5 Flash — genuinely cheaper and several times faster — is the more literal answer to “capable and cheap.” The takeaway is a mental model for every future launch: stop pricing AI by the token and start pricing it by the completed task.

I'm Dan. AI moves too fast to keep up with, so I built my own stack of AI tools to research, analyse, verify and illustrate the questions I can't stop thinking about — mostly to learn it myself, and I share what I find. AI-assisted, fact-checked, worth a second look.

Follow Dan’s AI Intel in whatever app you're listening in — it's free, and a follow is the single biggest thing that helps a small independent show grow. One quick thing, from Dan — I make this show mostly to keep up with AI myself, and I'd love to make it better. If there's something you'd push back on or want me to go deeper on, tell me: [email protected] — I read every one.