Loading…
Loading…
How many days your $100, $200 or $500 API credit lasts
Free, Pro and Enterprise plans get no API credit. New subscribers can claim after 7 days on an eligible plan, in claude.ai under Settings, then Billing.
System prompt, tools and context, cached or not. Haiku 5.5 reprices every run over 100,000.
Share of input billed at the cache-read price. Cache writes are not included.
Every API-key call in the linked Console org draws on the same balance.
No rollover. Whatever is left expires at the end of the billing cycle.
Percent of runs on each model. List prices per 1M tokens, checked 13 Oct 2026.
Fable 5.1
$10 / $50
$0.30per run
666runs for $200
Opus 5.5
$4 / $20
$0.12per run
1,666runs for $200
Sonnet 5.5
$2 / $10
$0.06per run
3,333runs for $200
Haiku 5.5
$0.10 / $0.50
$0.003per run
66,666runs for $200
50 runs a day · 20K in / 2K out · 100% Sonnet 5.5
$110of $200 expires unused
The credit resets on before you spend it, so $110 expires unused. Credits do not roll over. You have room for 111 runs a day, or move cheap batch work onto the balance in the last week of the cycle.
45% of the credit gets used before reset.
The credit pays for API usage on the Claude Platform. It does not change your usage limits in Claude or Claude Code.
claude -p and the Agent SDK when you run them yourself with an API key from the linked Console org.claude -p and Agent SDK usage when signed in with your Claude plan. Interactive Claude Code in the terminal, IDE, desktop or web.-p. Anthropic does not say in the same sentence which limit they draw on.The catch: read together, Anthropic’s two Help Center articles make the credential the rule. An API key from the linked org draws credits; plan sign-in draws plan limits. Neither page states that in one sentence, so test it: run one cheap claude -p call with the key and watch the Promotional credits balance in the Console move. Every key and workspace in the linked org shares one balance, so give each project a workspace with a spend limit.
Prices are Anthropic list prices per 1M tokens, checked on 13 Oct 2026: Fable 5.1 $10 / $50, Opus 5.5 $4 / $20, Sonnet 5.5 $2 / $10, Haiku 5.5 $0.10 / $0.50 up to 100K prompt tokens and $0.50 / $2.50 above. Cache reads: $0.25, $0.20, $0.10 and $0.01. Excluded: cache-write fees, the 50% Message Batches discount, tool fees and taxes. Rates and coverage rules change, so re-check Anthropic’s pages before you commit a budget. Pricing a multi-step loop? Use the AI agent cost simulator or the AI model cost calculator.
Run your projects on Claude Code?
We packaged the .claude config that runs this site — 26 specialist agents, 14 workflow skills, and 6 rule files from 31 real production incidents. From $9.
Get the Claude Code Production Pack — $29Since 7 October 2026 Anthropic gives Claude Max 5x subscribers $100 a month in API credits, Max 20x $200 and Team plans up to $500 pooled, and none of it rolls over. This planner turns a plan, a model mix and the size of one run into cost per run, runs the credit buys, days it lasts at your daily volume and the date it runs out. It flags credit that would expire unused and lays out what draws on credits versus plan limits, following our post Claude Max API Credits: What $100 to $500 a Month Buys.
A run is one API call. It costs its input tokens at the input price plus its output tokens at the output price, with prices per 1M tokens. The cache-read share of input is billed at the cache-read price instead: $0.25 per 1M on Fable 5.1, $0.20 on Opus 5.5, $0.10 on Sonnet 5.5 and $0.01 on Haiku 5.5. Haiku 5.5 is the only model with a tier: a prompt over 100,000 input tokens, cache reads included, moves the whole run to $0.50 / $2.50.
The blended cost per run weights each model by its share of runs, and the shares must add up to 100%. Runs the credit buys is the credit divided by the blended cost, rounded down. Days it lasts is those runs divided by runs per day, rounded down, and the run-out date counts forward from today in your browser’s time zone.
Credits expire at reset, so the planner compares spend before reset with the credit. Run out first and it shows the days you go without credit; reach the reset first and it shows the dollars that expire unused and the runs per day that would spend the whole balance.
Worked example: Max 20x, 100% Sonnet 5.5, 20,000 input and 2,000 output tokens, no caching. Input costs $0.04 and output $0.02, so a run is $0.06 and $200 buys 3,333 runs, which is 66 days at 50 runs a day.
A Max 20x subscriber running nightly claude -p jobs with an API key, checking whether $200 covers the month at 50 runs a day on Sonnet 5.5 or runs dry in week three.
A team lead sizing Team seats, who sees that five Premium seats already hit the $500 pool cap and a sixth adds no credit.
A developer routing a pipeline between Opus 5.5 for planning and Haiku 5.5 for triage, and checking which model takes most of the spend.
An agent builder whose context grows past 100K tokens, measuring what the Haiku 5.5 price cliff does to the runway before trimming history.
Scope note: Prices are Anthropic list prices checked on 13 October 2026. The planner leaves out cache-write fees, the 50% Message Batches discount, tool fees and taxes. The Haiku 5.5 cache-read rate above 100K ($0.05) is derived from the fivefold tier step rather than quoted. Whether headless Claude Code draws on credits depends on how you authenticate, so test one call before you budget.
Pick your plan: Max 5x ($100), Max 20x ($200), Team (enter Standard and Premium seats, pooled up to $500) or a custom monthly amount.
Describe one run: choose Chat turn (4K/1K), Agent task (20K/2K) or Long context (150K/3K), or type input tokens, output tokens and the cache-read share.
Split your runs across Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5 so the shares add up to 100%.
Set how many runs you make per day and how many days remain until your credit resets.
Read the cost per run, runs the credit buys, days it lasts and the run-out date, check the credits-vs-plan-limits panel, then share the card.
About the Claude Max API Credits Planner
Since 7 October 2026, Max 5x includes $100 a month and Max 20x $200. Team plans get $20 per Standard seat and $100 per Premium seat, pooled into one balance capped at $500, so three Standard plus two Premium seats give $260. Free, Pro and Enterprise plans are not eligible, and new subscribers can claim after 7 days on an eligible plan.
It depends on the model and the size of each run. At 20,000 input and 2,000 output tokens with no caching, a Sonnet 5.5 run costs $0.06, so $200 buys 3,333 runs, or 66 days at 50 runs a day. The same run costs $0.30 on Fable 5.1 (666 runs) and $0.003 on Haiku 5.5 (66,666 runs). Runway past the billing cycle is wasted, because the credit does not roll over.
Interactive Claude Code in the terminal, IDE, desktop or web does not; it runs on plan limits. claude -p and the Agent SDK draw on the credits when you run them with an API key from the linked Console org, and on plan limits when you are signed in with your Claude plan. The Claude Code GitHub Action, IDE extensions and desktop app count as Claude Code usage, so the credits do not cover them, even with -p. Test with one cheap call and watch the Promotional credits balance in the Console.
No. They expire at the end of each billing cycle, or monthly on annual plans. If you cancel or downgrade, credits you already hold stay usable until they expire, but new ones stop. The planner shows how much of the credit would expire unused at your daily volume and how many runs a day would use all of it.
Haiku 5.5 is priced by prompt length: $0.10 input and $0.50 output per 1M tokens up to 100,000 prompt tokens, and $0.50 and $2.50 above. The prompt length counts all input tokens, including cache reads and cache writes, and each request is priced on its own. A 100,000-token prompt with a 2,000-token reply costs $0.0110; at 100,001 tokens it costs $0.0550. Pick the Long context preset to see it.
API requests stop until the next monthly credit arrives, unless the org holds purchased credits or has auto-reload on. Usage is never billed to your Claude plan. Every API key and workspace in the linked org draws on the same balance, so give each project its own workspace with a spend limit.
Score your system prompt for bloat and get a cut list
Open →DeveloperPrice an agent fleet per run, day and month on every major model at once
Open →DeveloperBuilder/critic agent prompts with brakes built in
Open →DeveloperFormat, validate & diff JSON — runs entirely in browser
Open →