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

# Claude Fable 5.2

> Use Anthropic Claude Fable 5.2 through AnyFast for demanding reasoning, long-horizon agentic coding, research, and document work.

Claude Fable 5.2 is Anthropic's latest Fable model for demanding reasoning and long-horizon agentic work. Use the AnyFast model ID `claude-fable-5-2` with the native Anthropic Messages API at `POST /v1/messages`.

## Model specifications

| Specification | Value |
| - | - |
| Model ID | `claude-fable-5-2` |
| Context window | 1M tokens |
| Maximum output | 128K tokens |
| Input → output | Text and images → text |
| Thinking | Adaptive, always on |
| Default effort | `high` |
| Effort levels | `low`, `medium`, `high`, `xhigh`, `max` |
| Reliable knowledge cutoff | June 2026 |

## Migrating from Fable 5.1

The upstream prefix-binding check for Fable 5.2 thinking blocks applies to every account. Preserve `system`, `tools`, and earlier turns when replaying thinking blocks; append new turns only. Do not assume earlier models can read Fable 5.2 thinking blocks. On AnyFast, requests replaying the original thinking blocks still returned `200` after changing the system prompt or an earlier user message. The upstream binding error was not reproduced. Responses do not establish whether blocks were dropped or rewritten; do not rely on this permissive behavior.

## Quick example

```bash cURL theme={null}
curl https://www.anyfast.ai/v1/messages \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-fable-5-2",
    "max_tokens": 4096,
    "output_config": { "effort": "high" },
    "messages": [
      {
        "role": "user",
        "content": "Review this database migration plan and identify its three highest-risk assumptions: copy a 2 TB database overnight, pause writes for five minutes, switch all clients at once, and retain the old database for rollback."
      }
    ]
  }'
```

```python Python theme={null}
import anthropic

client = anthropic.Anthropic(
    api_key="YOUR_API_KEY",
    base_url="https://www.anyfast.ai"
)

message = client.messages.create(
    model="claude-fable-5-2",
    max_tokens=4096,
    output_config={"effort": "high"},
    messages=[
        {
            "role": "user",
            "content": "Review this database migration plan and identify its three highest-risk assumptions: copy a 2 TB database overnight, pause writes for five minutes, switch all clients at once, and retain the old database for rollback."
        }
    ]
)

for block in message.content:
    if block.type == "text":
        print(block.text)
```

```python Streaming theme={null}
import anthropic

client = anthropic.Anthropic(
    api_key="YOUR_API_KEY",
    base_url="https://www.anyfast.ai"
)

with client.messages.stream(
    model="claude-fable-5-2",
    max_tokens=4096,
    output_config={"effort": "high"},
    messages=[
        {"role": "user", "content": "Create a phased rollout plan for a database migration."}
    ],
) as stream:
    for text in stream.text_stream:
        print(text, end="")
```

## Core parameters

| Parameter | Type | Required | Description |
| - | - | - | - |
| `model` | string | Yes | Must be `claude-fable-5-2`. |
| `messages` | array | Yes | Conversation turns with `user` or `assistant` roles. Text, images, and tool results are supported. |
| `max_tokens` | integer | Yes | Maximum output tokens, including thinking and visible text. Range: 1–128000. |
| `output_config.effort` | string | No | Controls thinking depth. Supports `low`, `medium`, `high`, `xhigh`, and `max`; default is `high`. |
| `thinking.type` | string | No | Omit it or set it to `adaptive`. Thinking cannot be disabled. |
| `tools` | array | No | Tool definitions available to the model. |
| `tool_choice.type` | string | No | Supports `auto` and `none`. Forced `any` or a named `tool` is not supported. |
| `stream` | boolean | No | Enables SSE streaming. Default: `false`. |
| `stop_sequences` | array | No | Custom sequences that stop generation. |

> [!NOTE]
>
> Adaptive thinking is always on. Omit `thinking` or send `{"type":"adaptive"}`. Both manual extended thinking with `budget_tokens` and `{"type":"disabled"}` return `400`. Use `output_config.effort` to control reasoning depth.

> [!WARNING]
>
> Production sampling compatibility differs from upstream restrictions: separate requests with `temperature=0.5`, `top_p=0.5`, and `top_k=1` each returned HTTP `200`. Acceptance does not establish sampling effects or whether the gateway ignores or rewrites these values. Omit these parameters in normal calls.

## Compatibility differences observed in testing

Production also accepted the invalid `output_config.effort="not-a-level"`. Do not depend on server-side rejection: validate against the documented official values in your client. Omitting effort succeeded, but the response does not identify the internal effort level.

A `tool_choice={"type":"none"}` request produced no tool call, but returned `content: []` with `stop_reason: "end_turn"`, leaving no displayable text. Do not assume every HTTP `200` contains text blocks. The cause is not established.

`max_tokens=1` produced `stop_reason="max_tokens"`; `max_tokens=0` returned `400`. A stop-sequence test returned `stop_reason="stop_sequence"` and the matched `stop_sequence`, excluding that sequence from the text. The maximum output boundary remains untested.

## Tool use

Keep `tool_choice` at `auto` (the default) or use `none`. Tell the model in the prompt when a tool applies and describe its input schema precisely. Do not use `any` or a named forced tool choice.

## Multi-turn conversations

When you return an assistant turn containing thinking blocks, send the entire turn back exactly as received. Treat the system prompt, tool definitions, and earlier messages as append-only. The upstream API documents a binding error after prefix edits; AnyFast did not reproduce that error in the tested system/history edits.

> [!NOTE]
>
> Anthropic released Claude Fable 5.2 on November 9, 2026. See the official [model overview](https://platform.claude.com/docs/en/models/fable-5-2/overview), [what's new](https://platform.claude.com/docs/en/models/fable-5-2/whats-new-fable-5-2), and [prompting guide](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/prompting-claude-fable-5-2).

> [card] [API Reference](https://docs.anyfast.ai/api-reference/model-api/anthropic/claude-fable-5-2.md)
>
> View request fields, response blocks, and the interactive API playground.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.