Reviews · JULY 1, 2026
Reviewed: Claude Sonnet 5 lands as the default model at $2/$10, closes most of the Opus 4.8 gap
Anthropic's June 30 Sonnet refresh scores 63.2% on agentic coding versus Opus 4.8's 69.2%, ships with an updated tokenizer that expands input by up to 1.35×, and takes over as the default model for every Free and Pro user.
Anthropic shipped Claude Sonnet 5 on June 30, promoted it to the default model on every Free and Pro plan, and priced it at $2 per million input tokens and $10 per million output tokens through August 31. After that, list moves to $3 and $15. Opus 4.8 remains at $5 and $25, which puts the mid-tier at roughly a 40% list discount to the flagship, and closes most of the analytical gap that used to justify the spread.
The benchmark story is where the pitch lives. Per TechCrunch, Sonnet 5 posts 63.2% on the agentic coding evaluation, against 69.2% for Opus 4.8 and 58.1% for Sonnet 4.6. That's a 5.1-point jump generation-over-generation and a 6-point deficit against the flagship. Anthropic calls it "the most agentic Sonnet model yet," and frames the product line as a dial rather than a ladder: "Between Sonnet 5 and Opus 4.8, users can adjust the effort level to find the right balance of cost and performance." Anthropic also claims Sonnet 5 slightly outperforms Opus 4.8 on an unnamed knowledge-work benchmark, which is the kind of asterisk that tends to matter more in a sales deck than in production.
The catch is the tokenizer. The new encoder maps the same input to between 1.0× and 1.35× the tokens Sonnet 4.6 would've used. Anthropic says the introductory pricing "is set so that the transition to Sonnet 5 is roughly cost-neutral," which is a careful sentence. VentureBeat is blunter: the expansion "could quietly erode the pricing advantage for certain workloads." It's the same class of edit Anthropic made with Opus 4.7, and operators who didn't re-baseline then are the ones most likely to miss it now.
TechCrunch also reports lower rates of hallucination, sycophancy, and prompt-injection compliance versus Sonnet 4.6. The system card places Sonnet 5 below Claude Mythos 5 on every automated AI R&D evaluation, notes it doesn't cross the CB-2 autonomy threshold, and characterizes alignment risk as "very low, though higher than for previous Sonnet models." Cyber safeguards ship enabled by default at the same level used for Opus 4.7 and 4.8.
Distribution is already broad. GitHub's changelog confirms general availability across Copilot's Pro, Pro+, Max, Business, and Enterprise tiers, including Zero Data Retention. Claude Code ships with claude-sonnet-5 as the API identifier, and Fable 5 returns globally July 1.
The read: Anthropic has moved the price/performance frontier down a tier, on purpose, on the eve of a widely reported IPO push. The frontier model still exists to be sold; it just no longer has to do the work of being the default.