Skip to content
Today
Sign in

BuildNot yet confirmed elsewhere1 publisher2 min readPublished

Anthropic's $200 Max 20x API credit buys 667 Fable 5.1 agent runs or 66,666 on Haiku 5.5

Anthropic gave Max 5x subscribers $100 a month in API credits from 7 October, Max 20x $200 and Team plans up to $500 pooled. The credit excludes interactive Claude Code and expires each cycle, so the model a team runs sets what it buys.

The Engineer · Build desk

How we use AISend a correction

Illustration accompanying Anthropic's $200 Max 20x API credit buys 667 Fable 5.1 agent runs or 66,666 on Haiku 5.5
Generated illustration
Fable 5.1 costs most per run of four models Cost of one reference run on each model at its per-million-token price, per the dev.to write-up.

Comparison of cost per reference run: Fable 5.1 $0.30, Opus 5.5 $0.12, Sonnet 5.5 $0.06 and Haiku 5.5 $0.003.

Fable 5.1 costs most per run of four models (Cost per reference run, per the dev.to write-up)
ItemValueClaim
Fable 5.1$0.3 per run11
Opus 5.5$0.12 per run12
Sonnet 5.5$0.06 per run12
Haiku 5.5$0.003 per run12

What happened

  • Each plan links to one Console organization, each organization takes credits from one plan, and the subscriber cannot change the link.
  • The playground is inside the credit, while extra usage in the Claude apps and Claude on Bedrock, Google Cloud or Microsoft Foundry are outside it.
  • At zero balance, API requests stop until the next monthly credit arrives, and usage is never charged to the plan.

Why it matters

  • cost At the reference task a Fable 5.1 run costs 100 times a Haiku 5.5 run, so the model a team picks sets how long a month's credit lasts.
  • constraint The org link cannot be changed by the subscriber and unspent balance expires at cycle end, so a wrong link or an idle month cannot be recovered.
  • contradiction Two Help Center articles describe claude -p coverage in different terms, so a team signed in with a plan login can burn plan limits while its API credit expires unused.
  • decision Teams must choose which credential and workspace each job runs under, and the write-up recommends a separate workspace with a spend limit for each project.

The per-model counts come from one reference task: 20,000 input tokens and 2,000 output tokens, with no caching and no batch discount [10]. On Fable 5.1, at $10 input and $50 output per million tokens, input costs $0.20 and output $0.10, so a run costs $0.30 [11]. Dividing $200 by $0.30 gives 666 full runs, which the write-up's summary rounds to roughly 667 [11][4].

The same task costs $0.12 on Opus 5.5, $0.06 on Sonnet 5.5 and $0.003 on Haiku 5.5 [12]. On $200 that is 1,666, 3,333 and 66,666 runs [12]. A Fable run costs 100 times a Haiku run [14].

Those counts transfer only to a workload shaped like the reference task. Input is about two-thirds of the Fable cost [15]. An agent that carries 40,000 tokens of context into each step would cost $0.50 a run and stretch $200 to 400 runs [16]. That figure is our hypothetical, not the write-up's. The Haiku row also holds only for prompts up to 100K, and the write-up flags a price cliff beyond that size [13].

Team credit is capped as well as pooled. At $100 a Premium seat, the $500 cap is reached at five Premium seats, or at 25 Standard seats at $20 each [17]. Seats beyond the cap add no credit [2].

Coverage turns on the credential, and that is where credit gets wasted. Per the dev.to write-up's reading of Anthropic's Help Center, claude -p and the Agent SDK draw credits when run with an API key from the linked org [18]. Signed in with a plan, they draw plan limits and leave the credits alone [18]. The write-up says Anthropic does not state that rule in one sentence on either page, and a second article lists claude -p among the things the credits "cover" [19]. Its suggested test is one cheap claude -p call with the key, then a look at whether the Promotional credits balance in the Console moves [19].

Every API key and workspace in the linked org draws from one balance [8]. A runaway loop in one workspace therefore draws down the credit every other project depends on. Simon Willison, who noted the change the day it surfaced, wrote that the credits "do not roll over" [3]. Anthropic's page also says the credits "don't change your usage limits in Claude or Claude Code" [7]. A request outside coverage, from an org holding only these credits, shows "Credit balance too low" in Claude Code [9].

What to watch

  • Whether Anthropic states the claude -p credential rule in a single sentence on its Help Center pages.
  • The result of a one-call claude -p test against the Promotional credits balance in a linked Console org.
  • Any change to the non-rollover rule or to the lock on the linked Console organization.

Clarity's read

What the record supports and how the coverage leans. The claims behind it follow.

Reality

Evidence62
Adoption
Insufficient
Hype gap+5
Incentives25
Confidence60
Why these scores

Claim ledger

Ranked by verification strength, evidence, and original report placement.

  1. [1]

    On 7 October 2026 Anthropic started giving Claude Max 5x subscribers $100 a month in API credits, Max 20x subscribers $200, and Team plans up to $500 pooled; unused credit disappears at the end of each billing cycle.

    ReportedSupportedSource: dev.to write-up of Anthropic's developer page2 sources— create a free account to open themView cited source
  2. [2]

    Team credits are $20 per Standard seat and $100 per Premium seat, pooled into one balance capped at $500. Anthropic's own example: three Standard seats plus two Premium seats gives $260 a month.

    ReportedSupportedSource: Anthropic developer page, as reported by dev.to2 sources— create a free account to open themView cited source
  3. [3]

    Simon Willison noted the change the same day it surfaced in the Haiku 5.5 launch discussion on Hacker News: Max 5x gets $100 a month, Max 20x gets $200, Team gets "up to $500, pooled across their users", and the credits "do not roll over".

    ReportedSupportedSource: Simon Willison, as quoted by dev.to2 sources— create a free account to open themView cited source

Sources

1 independent publisher whose own reporting we read for this story.

  1. dev.to

    1 article · October 13, 2026

    Claude Max API Credits: What $100 to $500 a Month Buys

Share your take

Let Clarity write the post for you.

Signed-in readers get a short post drafted on this story in the register they choose — narrative, analytical, or a direct position — editable to the last word before it goes anywhere. The share buttons at the top of this story work without an account.

Create a free account to composeSign in

Topics and entities

Follow any of these and your For You feed starts watching them — no settings page required.

Topics

Entities

Loading related stories