How Much Does Claude Fable 5 Cost? Fable 5.1 Official Rates and Tiers
The current Fable model is Claude Fable 5.1: the official API price is $10 input / MTok and $50 output / MTok, with $0.25 / MTok for cache reads and $12.50 / MTok for cache writes, and a 1M token context window. Fable 5's official API price is $10 input / MTok and $50 output / MTok. That's a different thing from a Claude.ai Pro/Max subscription, and different again from how you're actually billed for Fable 5 on QCode.
Prices change: on QCode, Claude, GPT and Chinese models are billed per token — live rates on /models.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Checked on 2026-10-20: Anthropic's official price list places Claude Fable 5.1 in the Latest models section, at $10 input / MTok and $50 output / MTok, with $0.25 / MTok for prompt cache reads and $12.50 / MTok for cache writes, and a 1M token context window. Claude Fable 5 is $10 input / MTok and $50 output / MTok, with no recorded change since its official launch on 2026-06-09. Note that this is the rate for calling the official API directly -- it is not the Claude.ai subscription fee, and it is not QCode's billing model.
Last checked: 2026-10-20 (Anthropic's official price list and models overview).
Official Price List and Current Model Rates
Fable 5 next to a few other current flagship-tier models covered on this site, by per-million-token rate and context window -- the rows below are the current lineup in Anthropic's official price list (Latest models), with input and output rates and context windows as published; cache read and write rates are in the note column.
| Model | Input price / million tokens | Output price / million tokens | Context window | Note |
|---|---|---|---|---|
| Claude Fable 5.1 | $10 per million tokens | $50 per million tokens | 1M token context window | Current Fable tier, released 2026-09-01; the previous Claude Fable 5 launched 2026-06-09 with the same $10 input and $50 output, cache read $1 / MTok |
| Claude Opus 5.5 | $4 per million tokens | $20 per million tokens | 1M token context window | The verified per-token rate is published instead of a guessed number: Anthropic's official price list gives the previous Claude Opus 4.8 $5 input and $25 output per MTok, cache read $0.50 / MTok; official advice is to start with the current Opus for most workloads |
| Claude Sonnet 5.5 | $2 per million tokens | $10 per million tokens | 1M token context window | Cache read $0.10 / MTok; the previous Claude Sonnet 5 was $2 input and $10 output per MTok, cache read $0.20 / MTok |
| GPT-6.1 Sol | $2 per million tokens (OpenAI's official rate) | $10 per million tokens (OpenAI's official rate) | 1,050,000 token context window | OpenAI's official rate: $2.00 input and $10.00 output per million tokens, $0.10 cached input and $2.50 cache writes — a verified per-token rate, not a number this page guessed. |
Official Pricing Details
Fable 5.1 Official Rate Breakdown
The current Claude Fable 5.1 is billed at $10 input / MTok and $50 output / MTok, with prompt cache reads at $0.25 / MTok, 5-minute cache writes at $12.50 / MTok and 1-hour cache writes at $20 / MTok; the context window is 1M tokens, the maximum output on the synchronous Messages API is 128K tokens, and the Claude API model id is claude-fable-5-1 (MTok is one million tokens). Anthropic positions Fable 5.1 as next-generation intelligence for long-running agents, and the official selection advice reads: "If you're unsure which model to use, start with Claude Opus 5.5 for most workloads." Calling the Anthropic API directly, Claude Fable 5.1 is billed at $10 input / $50 output per MTok, with a context window of up to 1M tokens. This is pure usage-based pricing with no subscription or plan discount folded in. The previous Claude Fable 5 has the same input and output rates and a cache read of $1 / MTok; Anthropic now lists it as a previous-generation model, released 2026-06-09, and recommends migrating to Fable 5.1.
Cache pricing: published by Anthropic (0.1x / 0.025x), read $0.25 / MTok
Many models discount cached input tokens (cache read). Anthropic's pricing page lists the Fable numbers outright: Claude Fable 5 = $1 / MTok and Claude Fable 5.1 = $0.25 / MTok, plus $12.50 / MTok for 5-minute cache writes and $20 / MTok for 1-hour writes, on a $10 / MTok base. The same page states it as multipliers: "Cache hits and refreshes on Claude Fable 5.1 and Claude Mythos 5.1 are priced at 0.025x the base input price. All other models use the standard 0.1x multiplier." The current lineup adds Opus 5.5 at $0.20 / MTok for a cache read, Sonnet 5.5 at $0.10 / MTok, and Haiku 5.5 at $0.01 / MTok for prompts up to 100K tokens and $0.05 / MTok above that. Cache write tokens are charged when content is first stored, and a later request that retrieves that content is charged at the read rate.
Claude.ai subscriptions (Pro/Max) aren't the same thing as API billing
Claude.ai Pro/Max is a subscription: a fixed monthly fee buys tiered access to models including Fable 5, capped by usage limits or rate limiting rather than metered per token. API billing, by contrast, is precisely metered on input/output tokens with no monthly cap and no fixed fee. The two pricing models follow different logic -- you can't divide a subscription's monthly fee by some assumed usage to back into an equivalent rate, and you can't multiply the API rate by a subscription's usage allowance to decide whether the subscription is worth it. They serve different usage patterns.
Official tiers, cache billing and how QCode bills
Official rates and cache read/write prices for the current models, how cost differs between light and heavy tasks, and how billing actually works on QCode.
Cost estimate at the official rates (the assumed input is this page's own figure, not measured data)
This is an arithmetic estimate built on the official rate, not measured usage data. Assume a mid-sized PR: roughly 50,000 input tokens (including code context) and roughly 8,000 output tokens -- those token counts are this page's own stated assumption, and the official rates they are priced with are $10 per MTok for input and $50 per MTok for output, so you can recompute the cost of your own numbers with the same two unit prices. Actual token counts vary widely with codebase size, context length, and task complexity, so don't treat this as a measured baseline.
Cross-vendor comparison and light vs heavy tasks: how the official rates land on a bill
OpenAI's current counterpart is GPT-6.1 Sol, model id gpt-6.1-sol: $2.00 input / MTok, $0.10 cached input / MTok, $2.50 cache writes / MTok and $10.00 output / MTok, with a context window of 1,050,000 tokens and a maximum output of 128,000 tokens; the official page states that a request with more than 272K input tokens is priced at 2x the input and cache rates and 1.5x the output rate for the whole request. The previous GPT 5.6 sol, still listed on this site, is $5 input and $30 output per million tokens, with a cache read of $0.50 / MTok and a 922K token context window. A few-hundred-token single-turn question and a large refactor spanning tens of thousands of tokens can differ sharply on Fable 5.1 -- the official price list has Fable 5.1 at $50 output per MTok against $10 input per MTok, so long outputs (generating a large block of code or a long document) have an outsized effect on the total cost. In practice, it helps to estimate a task's rough input/output scale before running it, avoid judging very different task sizes against the same mental budget, and consider splitting or summarizing large-context, long-output tasks before feeding them to the model.
Using Fable 5.1 on QCode: how the per-token billing works
QCode's billing model is not the same system as Anthropic's direct billing. When you use Fable 5.1 on QCode, what you see billed (billed per token, with the per-model rates listed on /models, where Claude Fable 5.1, Claude Opus 5.5, Claude Sonnet 5.5, Claude Haiku 5.5 and GPT-6.1 Sol all appear, for Pay-as-you-go Packs the models are the same as on personal plans, 5 concurrent requests) is not the same thing as the Anthropic official API rate or the Claude.ai subscription fee described above, and you can't back into your actual QCode cost from the official rate. For the exact billing rules and your current balance, check the dashboard after logging in. When a Pay-as-you-go Pack runs out, requests return HTTP 402; after expiry they return HTTP 401.
Pricing Timeline
The record of official rates and cache prices from Fable 5 on 2026-06-09 to Fable 5.1 on 2026-09-01.
Claude Fable 5 launches at an official $10 / $50 per million tokens
Anthropic launches Claude Fable 5 at an official API rate of $10 input and $50 output per MTok, with a 1M token context window. On 2026-09-01 Anthropic released Claude Fable 5.1, which keeps the same $10 input and $50 output per MTok and lowers the cache read to $0.25 / MTok, against $1 / MTok for Fable 5.
Primary Sources
Anthropic's official API pricing page and price list at claude.com/pricing; the official models overview at platform.claude.com/docs/en/models/overview; the Claude Fable 5.1 and Claude Fable 5 model pages; OpenAI's official GPT-6.1 Sol model page at developers.openai.com/api/docs/models/gpt-6.1-sol; QCode's internal model price reference, verified 2026-10-20.
FAQ
What is the official API price of Claude Fable 5.1?
$10 input / MTok and $50 output / MTok; prompt cache reads $0.25 / MTok, 5-minute cache writes $12.50 / MTok and 1-hour cache writes $20 / MTok. The context window is 1M tokens and the maximum output is 128K tokens. This has been the official rate since the 2026-06-09 launch, with no recorded change on the official price list.
What do cache reads and cache writes cost on Fable 5.1?
Anthropic publishes it: $1 / MTok for Claude Fable 5 and $0.25 / MTok for Claude Fable 5.1, plus $12.50 / MTok for 5-minute cache writes and $20 / MTok for 1-hour cache writes (pricing-page table, base input $10 / MTok, i.e. the standard 0.1x and the 0.025x tier for 5.1 / Mythos). Cache writes are charged when the content is first stored, and a later request that retrieves it is charged at the read rate.
Is a Claude.ai Pro/Max subscription cheaper than paying the API rate directly?
The two follow different pricing logic and can't be directly compared. A subscription is a fixed monthly fee for tiered access; API billing is precisely metered per token. Light, everyday conversational use may be more convenient on a subscription; heavy agentic or tool-calling workloads are often billed via the API rate instead, since that allows precise cost tracking and attribution. Which is cheaper depends on your actual usage pattern.
How does billing work when you use Fable 5.1 on QCode?
No, it's a different billing system. QCode bills per token, with the per-model rates listed on /models, and its billing mechanism differs from both paying Anthropic's API directly and from a Claude.ai subscription. For the exact rules, check the dashboard after logging in -- don't use the official rate listed on this page to estimate your actual QCode cost. You can't buy packs while you have an active personal plan; if the plan's quota isn't enough, buy a booster pack. Not ready for a monthly plan? Pay-as-you-go Packs start at $4 for a one-time quota you use within 7 days (top-ups start at $5).
Create a QCode account
One key across Claude, GPT and Chinese models, billed per token — live rates on /models.
Related Guides
Claude Fable 5: The Complete Guide
Capabilities, use cases, and how to access it
Fable 5 vs GPT-5.6 Sol
Official rates and context windows of the current and previous flagships
QCode Pricing Guide
Model prices and billing across the whole site
This site is not affiliated with Anthropic. Official pricing follows Anthropic's own announcements and may change; the unit prices and model tiers here come from Anthropic's official price list and OpenAI's official model pages. Cost estimates on this page are arithmetic models built on the official rate, with the token counts as this page's own stated assumption rather than measured data.