Claude Fable 5 price: when e make sense to use am
Claude Fable 5 na $10 per million input tokens and $50 per million output. See wetin the published rates mean for the real jobs wey you dey run.
How much Claude Fable 5 dey cost?
Claude Fable 5 dey cost $10 per million input tokens and $50 per million output tokens for Claude API. Na dis be the official rates wey Anthropic publish, wey dem read from the pricing page for 21 October 2026. The API model ID na claude-fable-5. E come dey available for everybody for 9 June 2026 for Claude API, Amazon Bedrock, Claude Platform for AWS, Google Cloud, and Microsoft Foundry. Anthropic don list am now as legacy model. The one wey follow am, Claude Fable 5.1 (claude-fable-5-1), dem release am for 1 September 2026 with the same input and output prices. The cache reads dey cost $0.25 per million tokens, wey be one-quarter of the Fable 5 rate. Fable 5 still dey available and dem no go retire am before 9 June 2027.
Di data wey dey behind dis chart
[
{
"label": "Claude Fable 5",
"input": "10",
"output": "50",
"cache_read": "1",
"batch_input": "5",
"batch_output": "25"
},
{
"label": "Claude Opus 5",
"input": "5",
"output": "25",
"cache_read": "0.50",
"batch_input": "2.50",
"batch_output": "12.50"
},
{
"label": "Claude Sonnet 5",
"input": "2",
"output": "10",
"cache_read": "0.20",
"batch_input": "1",
"batch_output": "5"
},
{
"label": "Claude Sonnet 4.6",
"input": "3",
"output": "15",
"cache_read": "0.30",
"batch_input": "1.50",
"batch_output": "7.50"
},
{
"label": "Claude Haiku 4.5",
"input": "1",
"output": "5",
"cache_read": "0.10",
"batch_input": "0.50",
"batch_output": "2.50"
}
]Make you look am like ladder. Fable 5 price be two times Claude Opus 5, five times Claude Sonnet 5, and ten times Claude Haiku 4.5. The ratio dey the same for both sides of the bill, because every model for that chart dey price output at exactly five times im own input rate. So if you know the input and output token split for one job, you fit convert one price go another by multiplication.
Sonnet 5 dey list at $2 and $10 per million tokens. Anthropic first sell that price as introductory rate until 31 August 2026, then dem make am the standard price and cancel the planned increase to $3 and $15. The one wey follow am, Claude Sonnet 5.5 (claude-sonnet-5-5), dem release am for 28 September 2026 with the same $2 and $10, and Anthropic don list Sonnet 5 now as legacy model. Sonnet 5.5 cache reads dey cost $0.10 per million tokens, wey be half of the Sonnet 5 rate. So the five times gap between Fable 5 and any Sonnet still dey, and you fit use the same Sonnet numbers for your budget for autumn.
Di discounts, plus di one multiplier wey dey increase price
Prompt caching na di biggest way to save money. For Fable 5, cache read dey cost one-tenth of di base input price, wey be $1 per million tokens. One-tenth na Anthropic default multiplier, e no be rule for every model. As of 21 October 2026, di pricing page set cache reads for 0.025 times base input for Fable 5.1 and Claude Mythos 5.1, and 0.05 times for Claude Opus 5.5 and Sonnet 5.5. Na so Fable 5.1 take reach $0.25 per million for di same $10 base, and Sonnet 5.5 reach $0.10 for $2 base. To write go di cache cost pass normal input token: 1.25 times base for di five-minute cache, and 2 times base for di one-hour cache. Five-minute cache dey pay for itself after one read, and one-hour cache after two reads. Dat calculation na di main reason why you suppose calculate where prompt caching go start to pay you back before you turn am on everywhere.
Di Batch API dey give 50% discount for both sides, so batched Fable 5 work dey cost $5 and $25 per million tokens. Batch requests no dey happen immediately, so dis one only dey help work wey no need answer sharp-sharp. Di two discounts dey stack, wey mean say cached, batched job go cheap for both sides.
Di 1M token context window dey included for standard rates for Claude 4.6 and models wey come after, wit one exception: Claude Haiku 5.5. No extra charge for long context for Fable 5, so 900k token request go bill for di same rate per token as 9k one. Haiku 5.5, wey dem release for 7 October 2026, dey use prompt length take price am. Once prompt pass 100,000 tokens, e input rate go rise from $0.10 go $0.50 per million tokens and e output rate from $0.50 go $2.50.
Di multiplier wey dey increase bill na data residency. If you set inference_geo: "us" to keep inference inside United States, e go apply 1.1 times multiplier to every token category, including cache reads and cache writes. Make you leave di default global routing unless say contract talk say make you change am.
One billing rule dey specific to dis model family. Fable 5 get safety classifiers wey fit decline request. When dat one happen, Messages API go return stop_reason: "refusal" as successful HTTP 200 response instead of error. Whether you go pay for refusal wey come before any output depend on e stop_details.category. As of 16 October 2026, refusals for bio, frontier_llm and reasoning_extraction categories dey bill like any oda request, for di rates of di model wey run am. Refusal wey come before any output for any oda category, or wit null category, no dey bill. Anthropic talk say dat list fit change, so check di category table for dem refusals documentation instead of to assume say refusal dey free. If you retry di same prompt for another Claude model, fallback credit go refund di prompt cache cost of di switch, so you no go pay to warm di same cache twice.
Which plans get Claude Fable 5?
As of 31 October 2026, di help centre article wey Anthropic write about Fable models talk say Fable 5 and Fable 5.1 dey available for all paid plans (Pro, Max, Ultra, Team and Enterprise). E still talk say Fable 5 no dey available for di Free plan, and di plan comparison table for Anthropic pricing page show "No" for di Fable row for Free. Pro na di cheapest seat for dat list, so how much Pro cost and where di usage limits go stop you show di minimum money wey you go pay to get access to Fable. If na only because of Fable you wan upgrade, first check whether to pay for Pro seat based on how you dey use am because small Fable allowance no really worth di subscription money on its own. For developers, e dey generally available for Claude API and through di cloud marketplaces wey we list up so. Anthropic API pricing documentation, wey dem check on 21 October 2026, answer "Are there free tiers or trials?" wit signup credit: "New users receive a small amount of free credits to test the API." Dem no name any amount, and for Fable 5 rates, small credit fit only do test, e no fit do serious work, so read wetin di free route to Claude API actually cover before you plan your work around am. Teams wey wan evaluate for enterprise fit ask Anthropic sales team about extended trials. Paid seat fit now come wit API credit too. Monthly Claude Platform credits, wey dem introduce on 7 October 2026, now dey give Max subscribers $100 per month, Ultra subscribers $200, and Team plans up to $500 per month wey dem pool across seats. Di credit dey work for any available Claude model, Fable 5 included. Free, Pro and Enterprise plans no qualify, and unused credit dey expire for di end of each billing cycle, or monthly for annual plans, and e no dey roll over.
To dey available for a plan no mean say e no get limit. Subscription no dey metered like API, because Pro seat no get any published token quota at all but e get five-hour session limit and separate weekly limit instead, so nothing for a plan dey convert straight to di per million figures wey dey up so. Di five-hour session dey start wit your first message and e dey reset fully five hours later. Di weekly limit dey reset for fixed time every week wey dem assign to your account, e no matter when you start to use Claude or when your subscription start. Usage dey depend on how long your conversation be, how e complex, di model, di features wey you use and di effort level wey you select. Di plan comparison table for Anthropic pricing page dey lock Fable behind a share of weekly usage limits for some seats and behind usage credits for others, and dat arrangement change pass one time during July 2026. Anthropic own statement on redeploying Fable 5 include di model for up to 50% of weekly usage limits through 7 July 2026 for Pro, Max, Team and select Enterprise plans, wit usage credits after dat date. Di help centre now record say di promotion end on 19 July 2026. As of 31 October 2026, Max and Ultra seats, premium Team seats and premium seats for seat-based Enterprise plans fit spend up to 50% of dia weekly usage limits on Fable models wit no extra cost. Dat share dey come out from di normal weekly limit and e no dey add to am. Pro, standard Team and standard seat-based Enterprise seats dey run Fable on pay-as-you-go usage credits from di start.
So dis page no go print per-plan limit for messages or tokens, and even dat 50% share na wetin di help centre talk for di day wey dem check am. Di 50% allowance wey dem offer from 1 July run through 19 July, pass two weeks, and eligible premium plans keep am after dat. Open di row wey dem label Fable for di plan comparison table for di day wey you decide, and treat any number wey you see for news article as outdated. If you take seat for Fable access and di allowance come dey smaller pass wetin di table make you expect, stepping back down before di next renewal go leave di month wey you don already pay for intact.
Wetin dey stable na di API rate. For API, Fable 5 dey bill at di prices for di chart up so. For subscription seat, once you pass di included share, you go either pay wit usage credits or you stop. Usage-based Enterprise plans make dat structure clear, because di Enterprise seat fee dey buy access while di tokens themselves still dey meter at API rates on top of am. Premium seats for seat-based Enterprise plans na di exception, wit Fable included inside di 50% share wey we describe up so. Dat one make di price list di number wey you suppose plan against, which na di same conclusion wey you go reach when you compare API billing against a subscription for any other Claude model. If you never pick a tier, start from which Claude plan match your usage. If dat seat go be your everyday assistant instead of just way to enter Fable, e worth am to check how Claude plans price compare to ChatGPT own before you commit to one.
Wey your bill dey pass wetin price ratio suggest
Two documented mechanisms dey make Fable 5 cost pass wetin straight price comparison imply.
The first one na tokenizer. Fable 5 dey use the tokenizer wey Claude Opus 4.7 introduce. Compared with models wey release before Opus 4.7, the same text dey produce roughly 30% more tokens, and the exact increase depend on the content. Haiku 4.5 come before that tokenizer, so the ten times gap for the price list dey closer to thirteen times when you feed both models the same text block. Sonnet 5 dey use the newer tokenizer, so Fable against Sonnet na fair per token comparison. Fable against Haiku no be fair comparison.
The second one na thinking. Adaptive thinking dey always on for Fable 5, and thinking: {"type": "disabled"} no dey supported. Dem dey bill thinking tokens as output tokens, with the full output rate. This model no dey return the raw chain of thought, and thinking.display default to "omitted", meaning say short answer wey user fit see fit carry plenty billed output count. Anthropic documentation talk am clearly: the billed output token count no match the visible token count for the response.
Read the breakdown instead make you guess. The usage.output_tokens_details.thinking_tokens field report how many billed output tokens go for reasoning:
{
"usage": {
"input_tokens": 25,
"output_tokens": 348,
"output_tokens_details": {
"thinking_tokens": 312
}
}
}For that example, wey come from Anthropic thinking documentation, 312 of the 348 billed output tokens na reasoning wey nobody ever see. When you stream, this breakdown dey arrive only for the final message_delta event, so client wey stop to read after the last text chunk no go ever record am.
The control na effort, wey you set for output_config.effort, with levels low, medium, high (the default), xhigh and max.
{
"model": "claude-fable-5",
"max_tokens": 32000,
"output_config": { "effort": "medium" },
"messages": [{ "role": "user", "content": "..." }]
}Anthropic guidance for this model na make you start with high, move up to xhigh only for work wey capability matter pass, and step down to medium or low for routine work. The reason na say lower effort for Fable 5 often beat xhigh for older models. Two traps dey here. If you change effort between requests, your prompt cache go become invalid, because the resolved effort value dey render inside the prompt. So choose one level for each workload and keep am constant. Fable 5.1 add per message effort change for beta, and e dey preserve the cache, but Fable 5 no support am. And max_tokens na hard limit for total output, including thinking and response text, so cap wey you size for answer without thinking fit truncate the response. If you see stop_reason: "max_tokens", either raise the cap or lower the effort.
Cost per completed task, no be cost per token
Di data wey dey behind dis chart
[
{
"label": "Short answer (5k in, 1k out)",
"fable_5": "0.10",
"opus_5": "0.05",
"sonnet_5": "0.02",
"haiku_4_5": "0.01"
},
{
"label": "Coding session (300k in, 40k out)",
"fable_5": "5.00",
"opus_5": "2.50",
"sonnet_5": "1.00",
"haiku_4_5": "0.50"
},
{
"label": "Long agent run (2M in, 400k out)",
"fable_5": "40.00",
"opus_5": "20.00",
"sonnet_5": "8.00",
"haiku_4_5": "4.00"
}
]Those 3 rows na arithmetic, dem no be benchmark. Dem don publish the rates. The token volumes na assumptions wey you suppose replace with figures from your own usage data. Coding session wey resemble the middle row go cost $5.00 for Fable 5, compared with $1.00 for Sonnet 5. The long run go cost $40.00, compared with $8.00. The Haiku column for that last row na arithmetic only: Haiku 4.5 get 200k token context window, so every request must fit inside that limit.
Cost per token no be the correct unit for decision. Cost per completed task na cost per attempt multiplied by number of attempts. With today's rates, one Fable 5 attempt cost the same as five Sonnet 5 attempts, so retry count alone almost never justify the upgrade. Sonnet must fail more than four times out of five before the cheaper path lose on tokens.
What justify the choice na everything wey the token bill no include. The gap between the two long runs for the chart smaller than one hour of engineer time for most markets. If the expensive model save one hour of review, or prevent one bad migration wey you later need undo, e don pay for itself through a bill wey no show the saving. Do the comparison with money per accepted result, and include your own time for the total. Know wetin one million tokens actually buy go make that estimate much less abstract.
Jobs wey Claude Fable 5 fit justify im price
Long-running agent work wey wrong step fit cost plenty. Fable 5 dey made for work wey fit take hours: e get 1M token context window, up to 128k output tokens for each request, and an xhigh effort level wey target tasks wey dey run pass thirty minutes with token budgets wey reach millions. Compare the run cost with the cleanup work after agent take wrong branch for step 40, and nobody notice am until step 300.
Migration or audit wey you go run once across big codebase. One-off work no get enough volume to spread the cost, so the per-token premium go enter one invoice, and the whole job fit stay inside one context window. If e fit run asynchronously, Batch API go cut the cost by half to $25 for each million output tokens.
Task wey cheaper model don fail twice already. The two failed attempts plus your debugging time cost pass one Fable attempt at the volumes for the chart above. Escalation na the cheaper branch, and na wetin model ladder dey for. If you still dey choose between tiers generally, the capability comparison between the Claude model tiers dey answer different question from the one wey this page dey answer.
Three jobs wey Sonnet 5 or Haiku 4.5 be the correct answer
High volume classification and extraction. Dem dey grade these jobs based on throughput and unit cost, and the quality ceiling no high reach where you go need expensive model. If Haiku 4.5 fit do the task correctly, Fable 5 wey cost ten times more no go give you any extra value wey you fit measure. Haiku 4.5 na legacy model now. Anthropic promise say dem no go retire am before 15 October 2026, and as of 21 October 2026, their deprecations page still list am as active without any retirement date. Anthropic dey give at least 60 days' notice before dem retire any model wey dem don release for public. Anthropic docs point to Claude Haiku 5.5 as the model wey you suppose migrate go, so read how much Haiku 5.5 cost and where the long prompt pricing start before you build new pipeline on top the old one.
Interactive work where latency na the main product. Anthropic current model table rate Fable 5.1, wey be Fable 5 successor for the same price, as slower; e get the slowest rating for the lineup. You no fit turn off thinking for any of the Fable models. Chat box or editor completion dey feel worse for the more capable model, because users dey notice the wait time before dem even notice the quality.
Anything wey depend on events after January 2026. Fable 5 reliable knowledge cutoff na January 2026. Opus 5 own na May 2026. The more expensive model na the one wey no too current, so for questions about recent things, you dey pay double for older knowledge unless you give am search or fetch tools. Fable 5.1 move the Fable cutoff go June 2026.
One constraint dey wey no concern cost at all. Fable 5 na Covered Model, so e require 30 day data retention and e no dey available under zero data retention unless Anthropic expressly authorize am. If your agreement require zero retention, Fable 5 go hard to use but e no mean say e impossible. Organisation wey get ZDR arrangement fit turn on 30 day retention for one workspace and run Fable there, while their other workspaces still keep zero retention. Anthropic 1 September 2026 Enterprise Frontier Safeguards announcement also talk say eligible customers go receive ZDR on Fable 5 and Fable 5.1 until those safeguards ready. Without that authorization or workspace wey get retention turned on, the API go reject Fable 5 requests, and no discount fit change that.
fable-method repos show wetin, and wetin dem no show
Practitioners don publish repos wey distil how Fable 5 approaches dey work into skills wey cheaper model fit follow. The fable-method repo describe one thinking skill, one orchestration loop, and one adversarial verifier wey dey run every claimed check again instead of just reading the agent own report. Its README talk am directly: "This na community distillation, e no be Anthropic artifact."
Treat am exactly like that. E show how people dey drive the model, but e no be documentation of how the model dey work. Anthropic no validate anything inside am, and several almost identical forks dey with different wording. Na wetin you go expect from folklore, no be from one spec.
The useful signal for cost decision be say the parts people find worth copying na procedural. Classify the task, name the check wey go prove say e don complete, gather evidence before you decide, then verify by observation instead of reading summary. If you give cheaper model explicit checklist and verification step, e fit close part of the gap. So run that experiment for your own evaluation set before you standardise on the expensive model. The result depend on your tasks, na why other people number no get much value for you.
Check your own numbers before you commit budget
- Read
usagefor every response, and logoutput_tokens_details.thinking_tokensseparately from wetin you dey see for output. Reasoning wey you no dey see na the line item wey dey surprise pipo. - Set
effortexplicitly instead of make you just accept the default, and make sure say e constant inside any conversation wey depend on prompt caching. - Put your stable prefix behind a cache breakpoint first. One cache read for one-tenth of base input price for Fable 5, and one-fortieth for Fable 5.1, dey save money pass wetin model downgrades fit do.
- Move anything wey no need immediate answer go the Batch API.
- Price the alternative honestly. Comparing Claude and ChatGPT API pricing for your own workload worth one afternoon before you commit long-term.
If the workload na agent wey dey run for your own server, the controls wey matter pass dey outside the model choice. Keeping agent costs under control on a VPS cover the loop limits and spend caps wey dey stop runaway session, and where Claude Code actually spends tokens explain the numbers wey you go see for your usage dashboard.
FAQ
How much does Claude Fable 5 cost per million tokens?
Claude Fable 5 price na $10 per million input tokens and $50 per million output tokens for Claude API, as we check am for Anthropic pricing page for 21 October 2026. Cache read for Fable 5 dey cost $1 per million tokens, wey be one tenth of the base input price. That ratio na only for Fable 5 and old models: for Fable 5.1, cache read dey cost $0.25 per million, wey be 2.5% of the base price. Batch API dey give 50% discount for both sides, so e dey cost $5 and $25 per million tokens for work wey no need immediate response. If you keep inference inside United States with the inference_geo parameter, e go add 1.1 times multiplier to every category.
Is Claude Fable 5 included in a Claude Pro subscription?
Anthropic help centre talk say Fable 5 dey available for all paid plans (Pro, Max, Ultra, Team and Enterprise) but e no dey for Free plan. To dey available for plan no mean say e dey included inside. As of 16 October 2026, Pro users dey access Fable through pay-as-you-go usage credits from the start. Max and Ultra seats, plus premium Team seats, fit spend up to 50% of their weekly usage limits on am without extra cost. That arrangement change pass once during July 2026, so make you read the row wey dem label Fable for the plan comparison table for Anthropic pricing page the day you wan decide, no be to just trust wetin you see for article.
Why is my Claude Fable 5 bill higher than the visible output suggests?
Adaptive thinking dey always on for Claude Fable 5, so thinking tokens dey bill as output tokens, and you no go ever see the raw chain of thought. With thinking.display for e default value of omitted, you go see empty thinking field but you still dey pay for every reasoning token wey dey behind. Read usage.output_tokens_details.thinking_tokens for the response to see the split, and note say when you dey stream, e go only show for the final message_delta event. Reduce the effort level if the ratio of reasoning to answer pass wetin the task need.
Should I use Claude Fable 5 or Claude Opus 5?
Claude Opus 5 cost half the price per token and e get more recent reliable knowledge cutoff, May 2026 compared to January 2026 for Fable 5. Start with Opus 5 and move up if the task fail there, or if one wrong answer cost pass the price difference. Fable 5 na the correct choice for long-term agent work where one bad branch fit cost you hours to clean up, and for one-off jobs where the premium go show for one invoice instead of every request forever. Both models na legacy now. Anthropic docs dey point new work to Claude Opus 5.5 at $4 and $20 per million tokens, or to Claude Fable 5.1 at the same price as Fable 5.