Anthropic quietly upgraded the two cheapest paid tiers of Claude

Opus 5.5 - a frontier-level AI model and the newest member of Anthropic's Claude products - was released this week, on September 22, 2026, and buried the news in the same release notes as a much louder product launch.

So if you, or your team, pays for Claude Pro or Claude Team Standard, the model doing that you use to do your heaviest lifting is significantly better today than the one that was doing it last week, which was already amazing. And to get it, you didn't have to click anything, upgrade anything, or pay a single cent more to get access to it.For some reason, that's the story that essentially nobody ran and I personally think it's pretty big news.So I will pick up the slack and fill you in as to what actually happened, what it's worth in real dollars, and what you should do about it THIS WEEK.

So what exactly changed on Claude's cheapest paid plans?

Excuse me, but the cheapest plans? That's right, the cheapest paid plans. Of course I'm referring to the $20/month PRO plan investment that, in most months, if leveraged properly can knock out more work while you sleep at night than many entire teams of people took 6 months to do just a couple of years ago, landed a new default model on September 22, 2026. Anthropic changed the default model on Claude Pro and Claude Team Standard from Sonnet to Opus — MATCHING what the considerably more expensive Claude Max, Team Premium, and Enterprise subscribers were already getting.

The release note is one line, sitting inside version 2.1.280 of Claude Code: "Changed the default model on Pro and Team Standard plans from Sonnet to Opus, matching Max, Team Premium, and Enterprise."

No price increase. No plan change. No settings to dig through. If you're on Pro ($20/month) or Team Standard ($25/seat), the model that answers you by default just moved up a full tier, automatically, mid-billing-cycle.

Why did nobody cover this?

Because it shipped in the shadow of a much bigger headline. That same day, Anthropic released Opus 5.5 — the new flagship model, and the plan-default change is what quietly made Opus 5.5 the default engine on the entry-level paid tiers too. The Opus 5.5 launch also landed the same week as OpenAI's GPT-6 Sol and Luna, and the entire tech press covered it as a pricing war between the two labs: better models, lower prices, competitive framing. TechCrunch, MacRumors, 9to5Mac, and CNBC all ran versions of the same story.

Now, a 'model launch' isn't something you necessarily act on. But a 'plan-tier default change that upgrades what you're already paying for'... quite possibly might be.

That's the distinction that got lost. "New model shipped" is a headline. "The model you already pay for just got better for free" is a Tuesday-afternoon decision that's actually worth five minutes of your time.

How much cheaper is the model you're now getting?

Opus 5.5 is priced at $4 per million input tokens and $20 per million output tokens, with cache reads at $0.20 per million.

According to Anthropic's own announcement, that makes it roughly 40% less expensive to run than Opus 5 on typical workloads — input and output tokens each dropped about 20%, and cache reads dropped around 60%.

But the exciting part is that Anthropic's own claim is that Opus 5.5 "performs at the level of Claude Fable 5.1 on most work" for a fraction of what Fable-tier performance used to cost.

You know... way, way back a couple of days ago.

Put plainly: the model itself got cheaper to run, and then Anthropic handed that cheaper, more capable model to its cheapest paying customers as the new default. That's not a limited time trial or some discount code. That's a quiet, permanent improvement to what you're already paying for.

Does this change the math on upgrading to Team Premium?

Well, at a minimum, it should make you re-check it. Claude Team Standard runs $25 per seat per month. Claude Team Premium runs $125 per seat per month — and there is a minimum of five seats on both of those plans, so 5X those prices, at a minimum. As much as anything, model access — specifically, getting your team onto Fable or Opus instead of Sonnet — has historically been part of the pitch for making that jump, and as much as the higher usage levels. But now, with the prices dropping, that means more usage, on a Fable level model, for $20/month. That's some math that might just make sense right now.

As of September 22, that specific justification is weaker. Team Standard now defaults to Opus 5.5 too. Team Premium almost certainly still earns its price on other grounds — higher usage limits, admin and governance controls, priority support are the usual suspects on tiered plans like this — but "you need Premium to get Fable" is no longer one of them.

If you've recommended a Team Premium upgrade to a client, or made that call for your own team, in the last few months specifically to get Fable access, that recommendation is worth revisiting. Not reversing on reflex — but revisiting, with current facts in hand.

What should you actually do about it this week?

Three things, in order of how fast you can do them:

• Check your plan tier. Log into your Claude account settings, or if you're inside Claude Code, run /status to see what plan and model you're actually on.
• Retest anything you previously shelved. If you or your team dismissed a task as "not worth running on the cheap tier" — a messier analysis, a longer draft, a more ambiguous research question — the calculus that made you dismiss it no longer holds. Run it again.
• Re-check any AI tooling advice you've given recently. If part of your reasoning for a client's plan upgrade was "you need the pricier tier to get the better model," that reasoning changed on September 22. Say so, and revise the recommendation if the new facts call for it.

Is there a catch?

The honest answer is: probably not a meaningful one, but check it yourself rather than taking our word for it. This update changes which model answers you by default — it doesn't appear to touch the usage limits (message or token caps) attached to each plan, based on Anthropic's own release note, which mentions only the model default. You can still manually switch to Sonnet with /model or through account settings if you'd rather trade some capability for speed or a lighter token footprint on a specific task. Nothing here forces you into Opus permanently — it just makes Opus the thing you get unless you choose otherwise.

Key Takeaways

• On September 22, 2026, Anthropic changed the default model on Claude Pro and Team Standard from Sonnet to Opus — at no extra cost and with no action required.
• Opus 5.5 itself is about 40% cheaper for Anthropic to run than Opus 5, per Anthropic's own announcement, and Anthropic says it performs at roughly Fable 5.1 levels on most work.
• This narrows, but doesn't erase, the model-access argument for upgrading from Team Standard ($25/seat) to Team Premium ($125/seat).
• Check your plan tier this week, retest work you previously wrote off as not worth the cheap tier, and revisit any recent AI-tooling advice that leaned on the old model gap.
• This story ran in nearly every outlet as "Opus 5.5 launches, prices drop in AI arms race with OpenAI." The one line that actually changes what you should do this week wasn't in any of those headlines.