> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tess.im/llms.txt
> Use this file to discover all available pages before exploring further.

# Claude Fable 5.1 (Anthropic)

Claude Fable 5.1 (Anthropic) is the successor to [Claude Fable 5](/en/claude-fable-5) in Tess — same input/output price, **cheaper cache reads**, and stronger **long-running agents**, **multistep research**, and **document / spreadsheet / slide** work. Released **1 September 2026**.

| **Model ID**<br /><br />`claude-fable-5-1`                                                   | **Context**<br /><br />1M input / 128K output | **Provider**<br /><br />Anthropic | **Released**<br /><br />1 Sep 2026   |
| :------------------------------------------------------------------------------------------- | :-------------------------------------------- | :-------------------------------- | :----------------------------------- |
| **Capabilities**<br /><br /><Icon icon="brain" /><Icon icon="image" /><Icon icon="wrench" /> | **Speed**<br /><br />Medium                   | **Cost**<br /><br />High          | **Intelligence**<br /><br />Frontier |

## What changed vs Claude Fable 5

|                           | Claude Fable 5                    | Claude Fable 5.1                                  |
| ------------------------- | --------------------------------- | ------------------------------------------------- |
| Focus                     | Frontier reasoning + long context | Long-horizon agents + knowledge work              |
| Context / max output      | 1M / 128K                         | 1M / 128K                                         |
| Thinking                  | Adaptive (always on)              | Adaptive (**always on**, default effort **high**) |
| Knowledge cutoff          | Jan 2026                          | **Jun 2026**                                      |
| Input → output            | Text + image → text               | Text + image → text                               |
| Tess credits / 100 tokens | 0.56 in / 2.80 out                | **0.56 in / 2.80 out** (same)                     |
| Provider cache read       | \$1 / MTok                        | **\$0.25 / MTok** (1/4 of Fable 5)                |

Native **reasoning**, **tools**, and **image input**. Anthropic’s public Fable line — reserve it for work that still falls short on lighter Claude models.

<Info>
  Anthropic classifies Fable 5.1 as a **Covered Model**: extra retention and safety review on the provider side. Follow the [Tess Privacy Policy](https://tess.im/page/privacy-policy) and [Anthropic’s retention guidance](https://trust.anthropic.com/resources?s=7ksqkied5hn0pocsj206m\&name=%5Banthropic%5D-security-and-privacy-design-of-anthropic-data-retention-and-review). Don’t send secrets, credentials, or regulated personal data unless the task needs them.
</Info>

## Pricing (Tess credits)

Values follow [Models and Costs](/en/models-and-cost) (credits per 100 tokens). Same input/output as Fable 5 — Anthropic did not change headline rates:

| Model            | Input / 100 tokens | Output / 100 tokens |
| ---------------- | ------------------ | ------------------- |
| Claude Fable 5.1 | 0.56               | 2.80                |

Long agent sessions that re-read a cached prefix are cheaper on the provider (cache reads at a quarter of Fable 5). Keep the thread on **5.1** to take advantage of that.

> **Screenshot placeholder — model picker:** Capture the chat model selector with **Claude Fable 5.1** selected.

## Ideal use cases in Tess

1. Long-running coding agents — features across a codebase, review, and performance work that spans many steps
2. Multistep research that has to keep the plan through tool calls
3. Knowledge work on documents, spreadsheets, and slides in one thread
4. Hard reasoning where Fable 5 (or a lighter Claude) still falls short
5. 1M-context sessions without switching models

<Tip>
  **Best practices**

  * Thinking **cannot be turned off**. Default effort is **high** — don’t use Fable 5.1 as a cheap extractor.
  * Prefer **Fable 5** only if you already have a working flow and don’t need the 5.1 agent/research gains; price per token is the same, cache is more expensive on 5.
  * Treat long chats as **append-only**. Editing earlier turns can drop or invalidate thinking blocks on the provider.
  * Cap response length in the prompt when you don’t need a long deliverable.
</Tip>

See also: [Claude Fable 5](/en/claude-fable-5) · [Models and Costs](/en/models-and-cost) · [Claude Fable 5.1 (Anthropic)](https://platform.claude.com/docs/en/models/fable-5-1/overview).