Claude Opus 5 Just Landed: What It Means If You Don’t Code

TL;DR: On 24 July 2026, Anthropic launched Claude Opus 5 — its fourth new model in under two months. It gets close to their top-tier Fable 5 model‘s intelligence at half the price, adds a low/medium/high “effort” dial so you can trade cost for capability per request, and is now the default model on Claude Max (and the strongest option on Claude Pro). If you build with Claude Code, this is the model doing the work behind the scenes from today — mostly good news, with one setting worth understanding.

Claude Opus 5 launch: what happened

On 24 July 2026, Anthropic announced Claude Opus 5, its fourth model release since Mythos 5, Fable 5 and Sonnet 5 landed over the preceding two months. The headline claim: Opus 5 “more than doubles Opus 4.8’s performance” on Anthropic’s own benchmark suite, while approaching Fable 5-level intelligence at roughly half the cost.

The concrete facts:

• Pricing stays the same as Opus 4.8 — $5 per million input tokens, $25 per million output tokens (standard mode). A faster “fast mode” runs about 2.5x quicker for double the price.
• New “effort” dial: you can now choose low, medium or high effort per task, as Fortune reported — letting you spend less on a quick fix and more on something that needs real reasoning.
• It’s now the default model on Claude Max, and the strongest model available on Claude Pro.
• Coding-tool builders reacted fast: Cursor said Opus 5 “delivers near Fable 5 intelligence at Opus speed and cost,” and Cognition (makers of Devin) reported it “approaches Fable-level performance at half the cost” on their coding benchmark.
• The launch drew heavy attention — a Hacker News thread hit 1,509 points and 840 comments, and it was covered same-day by Bloomberg and Fortune, both framing it as Anthropic addressing complaints about expensive AI bills.

Why it matters if you don’t code

If you build with Claude Code or use Claude Pro/Max day to day, this isn’t a tool you have to go and install — Opus 5 is quietly becoming the engine under the bonnet. A few things worth knowing:

• You’re not paying more. Same price as before, but for meaningfully more capable output — Anthropic’s own numbers claim more than double the performance on hard coding and knowledge-work tasks.
• The effort dial is the genuinely new, useful bit. Instead of guessing which model to pick (a headache we flagged just two days ago with Cursor’s new router), you get a direct dial on this model: turn it down for a quick tweak, turn it up when you need Claude to actually think something through. That’s a much easier decision than choosing between different model names.
• Less hand-holding, in theory. Anthropic says Opus 5 needs “less back and forth” and can check its own work and recover from errors without you stepping in — exactly the kind of friction non-developers feel most.
• One caveat: Anthropic itself still recommends Fable 5, its priciest model, for long, fully autonomous projects that need to run unattended for a while. Opus 5 is the everyday workhorse, not necessarily the right choice for your most ambitious unattended build.

How to try it

1. On Claude Max: you don’t need to do anything — Opus 5 is now your default model.
2. On Claude Pro: select Opus 5 from the model picker; it’s the strongest model available on your plan.
3. In Claude Code: check your model settings — Opus 5 (and its fast mode, billed through usage credits) should now be selectable for your sessions.
4. Try the effort dial. Set it to low for a small edit or quick question, high for something gnarly (a stubborn bug, a multi-step feature) — and notice the difference in both wait time and quality.
5. Building something long-running and fully autonomous? Stick with Fable 5 for that specific task, per Anthropic’s own guidance, and use Opus 5 for everything else.

Who should care (and who shouldn’t)

• Building a revenue app on Claude Code: worth a look today — likely better output at the same price, plus a cost dial to control spend.
• Building an internal tool for your business: the effort dial alone is a nice cost-control lever if your Claude usage has been creeping up.
• Making a website or landing page: low-stakes, but you’ll get the upgrade automatically on Max — no action needed.
• Learning to build: set the effort dial to medium and don’t overthink it; Anthropic designed this to need less fiddling, not more.
• Not sure which tool is even right for you: the quiz will point you at a tool before you worry about model settings — model choice only matters once you’ve picked your platform.

Our take

This is a genuinely useful release for this audience, more than most model launches are. The pricing hasn’t moved, the performance claims are backed by both Anthropic’s own numbers and two independent coding-tool makers (Cursor, Cognition) saying similar things in their own words — and the effort dial solves a real problem: most non-developers have no idea how to weigh “better AI” against “more expensive AI.” A three-way low/medium/high dial is something anyone can use without a manual.

We’d treat the “more than double the performance” headline claim as Anthropic’s own framing rather than independently verified — normal for a launch day — but the direction (cheaper power, an actual cost lever, less need to babysit it) is one non-developers should welcome. If you’re on Claude Code already, there’s nothing to lose in trying it today.

Not sure if Claude Code is even the right platform for you? Take the 60-second Vibe Coding Tool Finder quiz →

Start free with Claude Code → and see how it compares to OpenAI’s GPT-5.6, which launched a similar “which model do I need” trade-off two weeks ago.

Some links on this page are affiliate links. If you buy through them we may earn a commission at no extra cost to you.

FAQ

Do I need to pay more to use Claude Opus 5?

No. Standard pricing is unchanged from Opus 4.8 ($5 per million input tokens, $25 per million output tokens). A faster “fast mode” costs double but runs about 2.5x quicker — entirely optional.

What does the low/medium/high effort dial actually do?

It lets you choose how much the model “thinks” per task. Low effort is faster and cheaper for simple requests; high effort spends more time and budget reasoning through harder problems. It’s a built-in way to control cost without switching models entirely.

Should I use Opus 5 or Claude Fable 5 for my project?

Anthropic recommends Opus 5 for most everyday work — it’s now the default on Claude Max — but still points to Fable 5 for long, fully autonomous projects that need to run unattended. For most vibe-coding sessions with a human in the loop, Opus 5 is the sensible default.