Free Claude just got a lot better, and nobody sent you a memo.
On Monday, Anthropic launched Claude Sonnet 5.5. It's now the model free users get.
AI blogger Simon Willison put it simply: Anthropic now has "a much more capable free offering."
How much more capable? On a test of real office work (documents, spreadsheets, that kind of thing), Sonnet 5.5 scored 1844.
Opus 5.5, Anthropic's top model, scored 1846.
(Two points apart. That's a photo finish.)
The old Sonnet scored 1449. And the new one runs more than 30% faster.
So here's the short version: for most everyday work, the free model is now plenty, as long as you know two settings and one sentence.
Now a quick word from today's sponsor.
Stop Drowning In AI Information Overload
Your inbox is flooded with newsletters. Your feed is chaos. Somewhere in that noise are the insights that could transform your work—but who has time to find them?
The Deep View solves this. We read everything, analyze what matters, and deliver only the intelligence you need. No duplicate stories, no filler content, no wasted time. Just the essential AI developments that impact your industry, explained clearly and concisely.
Replace hours of scattered reading with five focused minutes. While others scramble to keep up, you'll stay ahead of developments that matter. 600,000+ professionals at top companies have already made this switch.
Right. Back to it.
What Sonnet 5.5 is great at
Anthropic's own description: "well-scoped everyday tasks," plus polished documents, slides and spreadsheets.
In real life that's emails, summaries, pulling the key points out of a PDF, brainstorming, a first draft of a deck.
It's also the first Sonnet to beat Pokémon Red working only from screenshots. (Useful for your job? No. Impressive? Very.)
When to switch to Opus
Anthropic says Opus 5.5 "remains clearly stronger at complex, open-ended work requiring sustained judgment."
In plain terms: long projects with lots of moving parts, tricky analysis where being right matters more than being fast, and big coding jobs.
One more reason. In an independent accuracy test by Artificial Analysis, Opus got 66% of factual questions right. Sonnet got 54%.
So when a fact really matters, use Opus or check the answer yourself.
To switch, click the model name next to the send button. Older models live under "More models." (What you can pick depends on your plan.)
Leave the effort setting alone (mostly)
If your plan shows it, click the model name, then Effort. You'll see Low, Medium, High, Extra high and Max. The apps start you on Medium.
The temptation is to crank it to Max. Don't.
Simon Willison tried Max. It "thought for 128,000 tokens" and then never gave him an answer.
Anthropic's own tests show Max scoring lower than Extra high on one coding test, because it made "extra edits beyond the task's scope."
And higher effort burns through your usage limits faster.
My rule: Medium for everyday work. High for anything you'd pay a person to think about. Max, basically never.
The one line that stops it overbuilding
Sonnet 5.5 is keen. Very keen.
Ask it for a few ideas and it might start building you the whole presentation. Anthropic's own guide admits it "can do more than you asked."
PCWorld spotted the catch: those unwanted extras "will eat into your Claude usage limits."
Anthropic's fix, word for word:
When the user asks for ideas, options or a plan, give them that and stop. Don't start building or changing anything until they say to go ahead.Open Settings and paste it into "Instructions for Claude." It applies to every chat from then on.
Make it stick
While you're in "Instructions for Claude," add two lines about you: what you do, and who you usually write for.
On the free plan your usage resets every five hours, so a model that stops when it's done is worth more than one that does extra.
That's one paste, and it carries into every chat.
Near-best answers pay off most when Claude has one clear job to do. If you run a business, that's exactly how Run On Claude works.
Every month, Claude takes one job off your plate. Right now that's your social posts and your customer emails, with a new job each month picked by member vote.
You set each job up once with a 20-minute guide. Claude writes the drafts, you read them and hit send.
If it hasn't saved you 5 hours in your first 30 days, you get your first month back.
Business owners only. $147 a month until we hit 50 members.
Talk tomorrow,
Zephyr
P.S. Try that one line for a day, then hit reply: did Claude calm down?