Claude Fable 5.2 API: pricing, examples, and Fable 5.1 comparison

Claude Fable 5.2 is the model Anthropic released on November 9, 2026 for demanding reasoning and long-running agent work. LaoZhang API has offered claude-fable-5-2 since November 10 at $8 input and $40 output per 1M tokens, the same as Anthropic’s list price and 20% below Fable 5.1.

Published: November 10, 2026
Last verified: November 10, 2026
Status: active; call it with a Claude Code or default group token

Key details

Item	Details
Model ID	`claude-fable-5-2`
Anthropic release	November 9, 2026
Available on LaoZhang API	November 10, 2026
Anthropic Messages endpoint	`POST https://api.laozhang.ai/v1/messages`, `Claude Code` token group
OpenAI-compatible endpoint	`POST https://api.laozhang.ai/v1/chat/completions`, `default` token group
Billing	Usage-based (per token), at Anthropic’s list price
Context window	1M tokens at one rate, with no long-context surcharge
Thinking	Always on; set the depth with `output_config.effort`, which defaults to `high`

Pricing

Prices from LaoZhang API’s public pricing as of November 10, 2026, in US dollars per 1M tokens. They match Anthropic’s pricing page.

Item	`claude-fable-5-2`
Input	$8
Output, including thinking	$40
Cache read	$0.2
Cache write	$10

Thinking tokens are billed as output, so a higher effort usually means more output tokens.
Check token usage and the charge for each request in call logs.
Current prices are in the model catalog.

Price comparison with Fable 5.1 and Opus 5.5

Model	Input	Output	Cache read
`claude-fable-5-2`	$8	$40	$0.2
`claude-fable-5-1`	$10	$50	$0.25
`claude-opus-5-5`	$4	$20	$0.2

Moving from Fable 5.1 to Fable 5.2 cuts input, output, and cache-read prices by 20%. The endpoints and token stay the same; only model changes.
Against Opus 5.5, Fable 5.2 costs twice as much for input and output, with the same cache-read price. Keep Opus 5.5 for work it already handles, and compare both on your own samples for multistep reasoning and long agent runs.
All three models have a 1M-token context window and are called the same way.

Make a request

Create an API key in Token management with the Usage first (recommended) or Usage-based billing mode:

For the Anthropic Messages format, including Claude Code and the Anthropic SDKs, choose the Claude Code token group.
For the OpenAI-compatible format, choose the default token group.

export LAOZHANG_CLAUDE_API_KEY="YOUR_CLAUDE_CODE_GROUP_KEY"
export LAOZHANG_API_KEY="YOUR_DEFAULT_GROUP_KEY"

For the Python example, run python -m pip install anthropic first.

Anthropic Messages
Python (Anthropic SDK)
OpenAI-compatible

curl https://api.laozhang.ai/v1/messages \
  -H "Authorization: Bearer $LAOZHANG_CLAUDE_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-fable-5-2",
    "max_tokens": 2048,
    "messages": [
      {"role": "user", "content": "Explain a database index in three sentences."}
    ]
  }'

import os
from anthropic import Anthropic

client = Anthropic(
    api_key=os.environ["LAOZHANG_CLAUDE_API_KEY"],
    base_url="https://api.laozhang.ai",
    timeout=300.0,
    max_retries=0,
)
message = client.messages.create(
    model="claude-fable-5-2",
    max_tokens=2048,
    messages=[{"role": "user", "content": "Explain a database index in three sentences."}],
)
for block in message.content:
    if block.type == "text":
        print(block.text)

curl https://api.laozhang.ai/v1/chat/completions \
  -H "Authorization: Bearer $LAOZHANG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-fable-5-2",
    "max_tokens": 2048,
    "messages": [
      {"role": "user", "content": "Explain a database index in three sentences."}
    ]
  }'

In the Anthropic Messages format, the answer is in the content[] blocks of type text. The first block can be a thinking block, so loop over the blocks by type instead of reading content[0].text.
In the OpenAI-compatible format, the answer is in choices[0].message.content.
max_tokens covers thinking tokens too. If it is too small the reply is cut off with stop_reason max_tokens; the examples use 2048.

Set the thinking depth

Thinking is always on for Fable 5.2. For faster, cheaper replies, add output_config.effort to a Messages request:

curl https://api.laozhang.ai/v1/messages \
  -H "Authorization: Bearer $LAOZHANG_CLAUDE_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-fable-5-2",
    "max_tokens": 2048,
    "output_config": {"effort": "low"},
    "messages": [
      {"role": "user", "content": "What is 17 times 23?"}
    ]
  }'

effort has five levels and defaults to high:

Level	What it does
`low`, `medium`	Less thinking and fewer output tokens
`high`	The default
`xhigh`, `max`	More thinking for the hardest tasks, more output tokens

Request differences

Anthropic sets three request rules for Fable 5.2. Check each one when you move from another Claude model:

Parameter	What to send
`thinking`	Omit it or send `{"type": "adaptive"}`; don’t send `disabled` or `budget_tokens`
`temperature`, `top_p`, `top_k`	Leave them out
`tool_choice`	Use `auto`; forcing a named tool returns 400

Streaming, tool use, prompt caching, and image input work as they do for other Claude models; see the Claude protocol reference. For how Claude models on LaoZhang API differ from Anthropic’s own API, see Claude model limits.

Who this affects

Applies to: projects that call Claude models through LaoZhang API, and anyone using Claude Code or the Anthropic SDKs.
Unchanged: no existing model changed. claude-fable-5-1, claude-fable-5, claude-opus-5-5, and the other Claude models keep working, and existing integrations need no changes.

FAQ

How much does the Claude Fable 5.2 API cost?

On LaoZhang API, claude-fable-5-2 costs $8 input, $40 output, $0.2 for cache reads, and $10 for cache writes per 1M tokens, the same as Anthropic’s list price (verified November 10, 2026). The rate is the same across the 1M-token context window.

Should I switch from Fable 5.1 to Fable 5.2?

Fable 5.2 costs 20% less for input, output, and cache reads, and switching only means changing model. First check the three parameters in Request differences, then compare quality and token usage on requests that represent your workload.

What is the Claude Fable 5.2 model ID?

The model ID is claude-fable-5-2 in both the Messages and the OpenAI-compatible format. There is no dated suffix.

Why does a request that forces a tool with tool_choice return 400?

Fable 5.2 can’t be forced to call a named tool. Change tool_choice to {"type": "auto"} ("auto" in the OpenAI-compatible format) and name the tool you want in the prompt.

How do I use Fable 5.2 in Claude Code?

Use a Claude Code group token and set the model ID in your configuration to claude-fable-5-2. The full setup is in Claude Code setup with LaoZhang API.

Anthropic pricing: list prices and cache multipliers for Fable 5.2.
Claude Fable 5.2 model page: release date, context window, thinking, and effort levels.
Claude API models and pricing: every Claude model on LaoZhang API.
Claude protocol: Anthropic Messages API
Model catalog
Latest announcements

​Key details

​Pricing

​Price comparison with Fable 5.1 and Opus 5.5

​Make a request

​Set the thinking depth

​Request differences

​Who this affects

​FAQ

​How much does the Claude Fable 5.2 API cost?

​Should I switch from Fable 5.1 to Fable 5.2?

​What is the Claude Fable 5.2 model ID?

​Why does a request that forces a tool with tool_choice return 400?

​How do I use Fable 5.2 in Claude Code?

​Sources and related links