Claude Sonnet 5 is the new default AI for millions of users, following a major launch at the end of June. Anthropic released the model on June 30 and made it the standard experience for all free and Pro users starting July 1. Here is what changed, what you can do with it, and how it fits into the broader AI landscape.
What Claude Sonnet 5 Is
The model represents a significant step forward in agentic capability. It is built to do more than answer questions. According to Build Fast With AI, Anthropic launched Claude Sonnet 5 as the most agentic Sonnet ever built, performing close to the flagship Opus 4.8 on many tasks, while at introductory pricing through August 31 it costs less than the prior Sonnet 4.6.
Anthropic’s own framing was direct: the model “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.”
Why Pricing Matters Here
The cost angle is a deliberate competitive move. Enterprise AI budgets became a real problem in early 2026. Enterprises recoiled from agentic AI bills in Q2 2026 as “tokenmaxxing” — running far more AI tokens than needed — burned through annual budgets in weeks.
Sonnet 5 at introductory pricing of $2 per million input tokens and $10 per million output tokens is Anthropic’s direct response: frontier-adjacent agentic capability at a price that keeps enterprise AI cost models viable.
How It Performs
The benchmark numbers give developers a concrete sense of what changed. On agentic coding, Sonnet 5 scores 63.2% on key evaluations compared to 58.1% for Sonnet 4.6. That is a meaningful improvement for developers using AI to write and maintain code.
A Zapier senior engineer described a two-part Salesforce automation that used to stall halfway through now completing end to end. These are production reliability improvements that translate directly to reduced human oversight costs per task.
The Competitive Context
Claude Sonnet 5 arrives during an intense period of AI model competition. Every major lab is racing to offer better capability at lower cost. Grok 4.3 from xAI is substantially cheaper for workloads that do not require top-tier agentic performance. Meanwhile, Google’s Gemini 3.5 Flash, launched at Google I/O, outperforms Gemini 3.1 Pro on agentic benchmarks while delivering output tokens four times faster.
The common trend is clear. The AI market is shifting from “which model is most impressive” to “which model delivers the most value per dollar.” Sonnet 5 is positioned squarely in that debate.
Who Gets It and How
All free and Pro Claude users gained access to Sonnet 5 as the default model from July 1. For developers building on the API, the introductory pricing runs through August 31.
For everyday users, the change is largely invisible, since the model simply replaces the prior default. For those doing more complex tasks like research, coding, or multi-step automation, the improvement in reliability and reasoning should be noticeable. As AI assistants become central to daily work, model releases like this one matter more than most people realise.
You may be interested in this article: ChatGPT Model Retirement: Which Versions Are Going Away and What to Use Instead.
