- Claude Sonnet 5 launched June 30 as the default model for Free and Pro plans, with agentic performance close to Opus 4.8, Anthropic's larger flagship.
- Introductory pricing of $2 per million input tokens and $10 per million output tokens runs through August 31, then rises to $3 and $15.
- The model ships with real-time cyber safeguards enabled by default, a first for the Sonnet line, three weeks after US export controls grounded Fable 5.
Sonnet 5 closes the gap with Opus 4.8 at less than half the price
Anthropic released Claude Sonnet 5 on June 30 and made it the default model for every Free and Pro user, according to the official announcement. The company describes it as the most agentic Sonnet yet: a midsize model that plans multi-step work, operates tools like browsers and terminals, and completes tasks that a few months ago required larger and more expensive models.
The economics are the story. Opus 4.8, Anthropic's flagship, costs $5 per million input tokens and $25 per million output tokens. Sonnet 5 launches at $2 and $10 through August 31, moving to $3 and $15 afterward. On Anthropic's published cost-performance curves for agentic search and computer use, Sonnet 5 at higher effort settings matches Opus 4.8 on some tasks while spending a fraction of the cost.
| Detail | Figure |
|---|---|
| Launch date | June 30, 2026, all plans and the Claude API |
| Introductory pricing (to Aug 31) | $2 input / $10 output per 1M tokens |
| Standard pricing (from Sep 1) | $3 input / $15 output per 1M tokens |
| Opus 4.8 reference pricing | $5 input / $25 output per 1M tokens |
| Predecessor | Sonnet 4.6, released February 2026 |
The release lands in a changed regulatory climate. Sonnet 5 is the first Sonnet model to ship with Anthropic's real-time cyber safeguards switched on from day one. The company reports the model cannot develop working software exploits in its Firefox vulnerability evaluation, scoring zero percent, and shows substantially weaker cyber capabilities than Opus 4.8. That caution follows directly from June, when a cybersecurity jailbreak triggered a 19-day government suspension of Claude Fable 5.
Claude Sonnet 5 is built to be the most agentic Sonnet model yet. It can make plans, use tools like browsers and terminals, and run autonomously at a level that, just a few months ago, required larger and more expensive models.Anthropic, Introducing Claude Sonnet 5, June 30, 2026
Why the cheapest capable agent matters more than the strongest one
Most production AI agents do not need frontier intelligence. They need reliability, tool use, and a cost per task low enough to run thousands of times a day. A model that approaches Opus 4.8 performance at 40 percent of its price resets the budget math for every team running agents at volume.
There is one caveat buried in the fine print. Sonnet 5 uses a new tokenizer that can map the same input to 1.0 to 1.35 times more tokens. Anthropic says the introductory pricing makes the transition roughly cost-neutral, which means the September price increase is the number enterprise buyers should model, not the launch price.
Sonnet 5 is not Anthropic's smartest model, and it does not need to be. By pushing near-flagship agentic capability into its default free tier while deliberately capping the model's cyber skills, Anthropic is making a bet that the next phase of AI competition is won on deployable, safe volume rather than benchmark ceilings.
Santage is committed to independent, transparent journalism. This article is produced in accordance with Santage's Editorial Standards and aims to provide accurate and timely information. Readers are encouraged to verify information independently.