Anthropic Made AI 3x Cheaper And Gave It Subagents. Coaches, The Floor Just Moved Again.

You set your prices last year.

Maybe the year before.

You priced against a market where AI was a clever assistant, not a workforce.

That market does not exist anymore as of today.

What Actually Shipped

Anthropic dropped Opus 4.8 on May 28. The headlines focused on coding benchmarks. The real story for coaches is in three lines further down the release notes.

One. A fast mode that is 2.5x faster and 3x cheaper than the prior version.

Two. A new dynamic workflow tool that lets Claude run multiple subagents at once. Not a single AI doing one task. A handful of AIs doing five tasks in parallel.

Three. Near Mythos-level alignment on "prosocial" traits like supporting user autonomy and acting in the user's best interest. Translation. The model now talks like it actually cares.

Why This Is A Pricing Event, Not A Tools Event

Every time a frontier lab cuts price and adds capability in the same week, something invisible happens in your sales calls.

Your prospect's mental anchor for "what AI work costs" drops a little.

They will not say it out loud. They will not even know they are doing it.

But when you quote five figures for content, systems, or strategy, a small voice in their head whispers "is this not... cheaper now?"

That voice just got louder today.

Intelligence keeps getting cheaper. Your fixed costs do not. The gap between what AI charges and what you charge is the part of your offer that has to do work the model cannot. Every release narrows the band of stuff you can charge for without justifying.

The Subagent Part Is The Bigger Story

Cheaper is the headline. Subagents are the shift.

The new dynamic workflow tool lets a single Claude session spin up multiple specialist sub-Claudes at once. One drafts. One critiques. One searches. One formats. They report back to a coordinator.

That is not "an AI assistant." That is the first version of a tiny org chart.

Until recently, the mental model was "AI does the task you would have given a junior." Today it is closer to "AI runs the team you would have hired."

For coaches, the implication is direct. The kind of work you used to package as "we will build this for you over four weeks" can be assembled in an afternoon by someone who knows how to point Claude at a goal.

The Mythos Part Almost Nobody Is Talking About

Anthropic also said it has made "swift progress" on safety safeguards that would let it release Mythos-class models to all customers. Soon.

Mythos is Anthropic's most powerful, currently restricted model. The one they have been holding back.

If that release happens in the coming weeks, the gap between "publicly available AI" and "frontier AI" closes again. The little moat indie coaches have ridden, the gap between what a buyer thinks AI can do and what it actually can do, narrows by the same amount.

None of this is bad news. It is timing data.

The Last 30 Days In One View

What This Resets For You

Open your last proposal.

Read the scope line by line.

Ask one question on every line. "Could someone who knows how to drive Opus 4.8 with subagents produce a reasonable version of this in a day?"

If the answer is yes, that line is no longer yours to charge for. Not at last year's number.

The lines where the answer is no, where the work requires a body, a witness, a container, a felt sense, a real relationship across months... those are the lines that hold their price.

Move your offer toward those lines. Faster than the model moves toward the others.

Your Move

Send a 12 word note to your last three clients today.

"Curious. What part of working with me would you not trust an AI to do?"

Whatever they say back is your moat for the next six months.

Charge for that. Quietly automate the rest.