AI status

Event Published Models and productsAnthropic

Anthropic releases Claude Fable 5.1 and cuts cache read pricing by 75%

Anthropic's Claude Fable 5.1 keeps its headline API price but cuts cache reads to $0.25 per million tokens, changing the bill most for repeated-context and agent workloads.

What happened

Published here

On 1 September 2026 Anthropic released Claude Fable 5.1 and Claude Mythos 5.1, the same underlying model at two safeguard levels. Fable 5.1 is generally available on the Claude API, claude.ai, Claude Code and the major clouds. Mythos 5.1 is restricted to vetted cybersecurity and life sciences organisations in the United States. API pricing holds at $10 per million input tokens and $50 per million output tokens, but cache reads drop 75% to $0.25 per million tokens. Anthropic puts the saving at roughly 25% for typical workloads and up to 45% for highly agentic ones. (Anthropic announcement, 1 September 2026)

Claude Fable 5.1 holds output at $50.00 and input at $10.00 per million while cache reads fall to $0.25.
The sticker prices held while the cache read line fell to a fortieth of them, and that is the line an agent loop hits again and again, drawn from Anthropic's own published prices as read on 5 September 2026. Diagram: AI Journey.

Open the diagram at full size

Read the full description of this diagram

Diagram in three parts. On the left, three bars drawn to scale for one repeated request: output tokens at $50.00 per million, fresh input tokens at $10.00 per million, and a petrol sliver for cache reads at $0.25 per million, labelled as the only line that moved. On the right, two outlined boxes for the same underlying model at two safeguard levels: Claude Fable 5.1, generally available on the Claude API, claude.ai, Claude Code and the major clouds, and Claude Mythos 5.1, restricted to vetted cybersecurity and life sciences organisations in the United States. Along the bottom, one short question, then an agent loop or a long conversation, then Anthropic's stated saving of about 25% for typical workloads and up to 45% for highly agentic ones.

The numbers

Cache reads, per million tokens

$0.25

A fortieth of the input price on the same card, which is why a repeated context moves a bill.

Input, per million tokens

$10

Held at the Fable 5 rate, so occasional use sees no change at all.

Output, per million tokens

$50

Also held, and the larger half of most bills.

Anthropic's own saving ceiling

45%

For highly agentic workloads. Roughly 25% for typical ones, on its own reckoning.

Read from Claude API pricing, where the per million token figures are published

Tracked on this site: our Claude API price table

What they are saying

Outlets read

  1. VentureBeatcheaper, and still not cheap

    Fable 5.1 nevertheless remains expensive relative to much of the broader market.

    read

  2. The Daily Briefthe rival sits inside the same price list

    Anthropic cut Claude Fable 5.1 cache reads 75% to $0.25/MTok, half of Opus 5's rate on a model with twice the base input price. But Opus 5 stays cheaper until cache reads pass a third of your bill, and Anthropic's own savings ceiling of 45% falls short of the 50% crossover.

    read

Our verdict

The readings above are arguing about different yardsticks: VentureBeat measures the model against what other vendors charge and calls it dear, while The Daily Brief measures one line of it against Anthropic's own flagship and says the cheaper choice has quietly swapped ends. Neither of them can settle it, because the number that decides is the share of your tokens that are repeats, and that figure lives on your invoice rather than on anybody's price page.

What it means for you

The sticker prices did not move, so anyone sending a few requests a day sees no change at all. The cut lands on cache reads, which is where the bill actually lives if you run agents, long conversations, or anything that sends the same context again and again. For those workloads this is a price cut of a quarter or more without touching a line of code. It is the first change on our Claude API pricing table since we began recording it, and Anthropic did not state any change to the consumer Claude plans.

In more detail

The cut is easy to miss, because it lands on the line almost nobody reads. Anthropic puts cache reads at $0.25 per million tokens after a reduction of 75%, which is another way of saying they stood at $1.00 before it. Set that against the input price on the same card, $10 per million tokens, and a cache read now costs a fortieth of what those same tokens cost the first time you sent them.

That ratio is the whole story. A cache read is what you pay when you send Claude context it has already seen, which is what happens on every turn of a long conversation and at every step an agent takes through the same standing instructions. Ask one short question and you never touch it. Run a long brief through step after step and you touch almost nothing else, which is why Anthropic’s own saving runs from roughly a quarter for ordinary use up to nearly half for the agentic kind.

Every figure above is Anthropic’s own published list price, read on the date in the sources, and every quotation above is marked as the outlet that wrote it. What this site adds is what comes next: the Claude row on our API price table now has a recorded starting point, so the next time any of these figures move, the move will be dated against this one instead of arriving as a number with no past. That is also the number the readings above are missing. Whether the cut is large depends on how much of your traffic is a repeat, and that ratio is the one figure neither a vendor page nor an outlet can know about you.

Sources

Sources read

Cite this

AI Journey, "Anthropic releases Claude Fable 5.1 and cuts cache read pricing by 75%", 6 Sep 2026, https://aijourney.space/updates/2026-09-06-claude-fable-5-1-cache-price-cut/ (source: Anthropic, verified 6 Sep 2026)

Next in the stream

More in Anthropic

All Anthropic news