Anthropic just released Claude Sonnet 5. Not as a premium product behind a paywall. Not as a developer-only upgrade. As the new default — for everyone, free, live on claude.ai right now.
That's actually unusual.
What changed?
If you've been using the free tier on claude.ai, you had a solid but unremarkable AI. Claude Sonnet 5 changes that. It's currently Anthropic's best freely accessible model — and the gap to their flagship Opus 4.8 (which targets developers and is priced accordingly) has narrowed considerably.
Put simply: you get more today without paying more.
What's new — no jargon
Three things stand out in everyday use:
A massive context window. Claude Sonnet 5 can currently handle up to one million tokens at once. A token is roughly a syllable. One million tokens equals several thick novels. In practice, this means you can throw in very long documents, carry on extended conversations, or get to the bottom of an entire stack of text without the model forgetting what you said at the beginning. Two caveats: that full million is the model's capability — it's mainly the paid plans that let you actually use it in full, while the free plan gives you a smaller window. And the more you throw in, the faster you burn through your free allowance (more on that below).
Fewer mistakes, sharper focus. Sonnet 5 handles multi-step tasks noticeably better than its predecessor. It hallucinates (invents facts) less often and stays on track with complex requests. Not a miracle cure — but a clear step up.
More autonomous. The model handles sequences of steps better without needing you to nudge it at every turn. For beginners, this matters less right now. If you ever start automating things, you'll feel the difference.
What can you try right now?
If you haven't used claude.ai yet: it's Anthropic's chat interface, similar to ChatGPT. No download, no installation — sign up and go.
Worth trying:
- Analyze long texts: Paste an entire PDF or article, ask for a summary, spot contradictions, or surface open questions.
- Translate with context: Not just a sentence, but a whole document. Style is preserved better than most alternatives.
- Write and revise: Emails, cover letters, short reports — Claude is direct and doesn't over-flatter.
- Complex one-off questions: Situations where a standard search fails because the answer depends on your specific context.
Is this for you?
Yes — if you:
- haven't used claude.ai much and are curious,
- assumed free AI models were never worth it,
- want to get something done longer, better, or more thoroughly than before.
Not really — if you:
- are happy with a different tool,
- mainly want to generate images or videos (that's not Claude's strength),
- are just starting out with any AI at all — then honestly, start anywhere, it doesn't matter.
What does it cost?
For users on claude.ai: currently free on the free plan, no credit card needed.
But free doesn't mean unlimited — and that's the catch. The free plan gives you a certain number of messages per time window, then you're on a break until your allowance refills. How much exactly depends on demand and on how long your requests are — big documents eat through it faster. For trying things out and occasional use, that's plenty. In everyday use it gets tight. And if you work with it intensively all day, you'll hit the wall reliably — that's what the paid plans are for. Otherwise nobody would have a reason to pay.
For developers accessing via the API: there's currently an introductory pricing phase in place. The exact numbers are on anthropic.com — they change, and I'd rather point you to the source than leave stale figures here.
Tl;dr: Claude Sonnet 5 is the best model you can get right now without paying extra — as long as you keep an eye on the free limits. Give it a try.
