Anthropic released Claude Sonnet 5 on June 30, 2026, and it is now the default version of Claude you get for free. It is also better at handling tasks that need multiple steps, like planning a trip or organizing your inbox. This is the kind of upgrade most people never notice, but you actually use it the next time you open Claude.
The Gist
- Claude Sonnet 5 is now the default model on the free version of Claude, so you get it without paying
- It is designed to handle multi-step tasks called “agents”, which means Claude can do a whole job on its own instead of just replying to one question
- Businesses pay 2 dollars per million words in and 10 dollars per million words out, which makes Claude cheaper than most rivals for these tasks
Have ChatGPT Recap This Article
ChatGPTWhat Actually Changed for You
If you open Claude today, whether on the web or in the app, you are talking to Sonnet 5. Anthropic, the company behind Claude, released this version on June 30, 2026, and made it the default model for free users. You do not have to click anything or pay anything to get it.
The word “model” is the name for the underlying AI. Every chatbot you know sits on top of a model. When Claude gets a new model, it is a bit like when your phone gets a new operating system. The way you use the app looks the same, but what happens under the hood is different.
The main upgrade with Sonnet 5 is what people call agent tasks. An agent, in AI terms, is when the assistant does not just answer one question and stop. It takes on a whole job that needs several steps. Booking a trip, comparing prices across sites, reading a long report and summarizing the important parts, all of those are agent tasks.
Before Sonnet 5, Claude would often get stuck halfway through this kind of task. It would forget what it was doing, or ask you a question mid-way, or just stop. Sonnet 5 is trained to finish these tasks more often, and to check its own work along the way. One engineer at the automation tool Zapier described the difference plainly: it used to break halfway, now it just gets through.
There is one honest thing to say about the free version. Free users always have usage limits. You can chat as much as you want in a short window, but if you use it heavily for a full day, you will hit a wall and have to wait. That is not a Sonnet 5 rule, that has always been how the free tier works.
Why the Price Matters Even If You Do Not Pay
Anthropic also announced the price businesses pay to use Sonnet 5 in their own apps. It costs 2 dollars per million tokens in and 10 dollars per million tokens out, at least until August 31. After that, the price goes up to 3 dollars in and 15 dollars out. A token, roughly, is a piece of a word. A short paragraph is about 100 tokens.
This might sound like something only developers care about. But it matters for regular users too. When a business builds an app on top of Sonnet 5, the cost of running the app depends on the token price. If the model is cheap, the app can be cheap. If the model is expensive, the app is either expensive or it caps how much you can use it.
Anthropic claims that Sonnet 5 is cheaper than the main rivals for agent tasks. That includes GPT-5.5 from OpenAI, Gemini 3.1 Pro from Google, and even Anthropic’s own more expensive model called Opus 4.8. The one model still cheaper is Gemini 3.5 Flash, but it is in a different performance category.
The practical result is that a lot of apps you use every day that quietly run Claude in the background just got cheaper to run. Over the next few months, expect more small tools built on top of Sonnet 5 because it is now the sensible default when a founder is picking which AI to plug in. there’s a rundown of the Sonnet 5 price cut on AI Noobies.
Anthropic is not alone in this pricing race. Just this week, we covered how Anthropic got the green light from the White House to sell Claude Mythos to 100 US companies, and how Claude Tag brings Claude into Slack as a real team assistant. Both are signals that the company is pushing hard to get its models used everywhere. The Sonnet 5 free rollout fits that same push.
Keep learning on AI Noobies:
- AI Labs Ban Each Other’s Tools at Big Tech
- Gemini Draws You for Free: Nano Banana Hits US
- Dawn Song Joins Meta: AI Agents Are the Next Milestone
What to Expect Over the Next Few Weeks
In the short term, if you already use Claude, the best thing to do is just try it on a task you would normally struggle to get done in one message. Ask it to plan something with multiple steps. Ask it to read a long document and pull out the key parts. See if it finishes the job without you having to nudge it.
If you have never really used Claude and you were curious, this is a good week to open the app. The free version is genuinely capable now. It is not a demo of a smaller model with the good stuff locked behind a paywall. The same model that businesses pay for is the one you use free, at least on the free tier’s usage limit.
In the medium term, expect Google and OpenAI to respond. When one lab drops the price on a strong model, the others usually match within weeks. That is good news for you. Over the next three months, the cost of running these AI tools should keep falling, which means more free features across all the apps you use.
There is also a shift in how these tools are used. Six months ago, most people used a chatbot to ask questions. Today, more and more people use it to get things done. Booking, planning, comparing, organizing. Sonnet 5 is built for that kind of use, not just chat. A related trend we already covered is how bots now run coding loops on their own while you sleep, which is the same idea applied to developer work.
The one thing to remember, whatever model you use, is that these tools still make mistakes. They are much better than they were a year ago, but they can still confidently give you wrong information. If you are using Sonnet 5 to plan or decide something that matters, double-check the important parts. The AI does the heavy lifting, you still hold the final call.
Stay tuned on AI Noobies.
About the Author
