> Markdown mirror of https://router.one/blog/claude-fable-5-2-api-guide for AI assistants and crawlers. Router One is a unified, OpenAI-compatible LLM API gateway.
> Published: 2026-11-11 · Author: Router One Team

# Claude Fable 5.2 API Guide: Model ID, Billing, Claude Code

_Claude Fable 5.2 on Router One: the anthropic/claude-fable-5.2 ID and claude-fable-5-2 alias, Anthropic's rules, wallet billing, Claude Code, first calls._

As of the 2026-11-11 catalog, Router One lists Claude Fable 5.2 as `anthropic/claude-fable-5.2`. The gateway also accepts Anthropic's own id, `claude-fable-5-2`, for the same catalog entry. Send `"model": "claude-fable-5-2"` to `https://api.router.one/v1/messages` with a Router One key — or point an OpenAI-compatible client at `https://api.router.one/v1` — and the request is routed, metered and logged like every other model. As of the 2026-11-11 plan response, no tier of the Pro, Max or Ultra plans lists Claude Fable 5.2, so every call bills per token to the wallet, plan or no plan, at the rates on the [Claude Fable 5.2 model page](https://router.one/models/claude-fable-5-2).

This guide covers what the catalog lists for the id, what Anthropic documents about the model on its own API, how to choose between Claude Fable 5.2, Claude Fable 5.1 and Claude Opus 5.5 on Router One, how billing and plans apply, how Claude Code selects it, the first request on each endpoint, and which clients can call it. Catalog and plan observations and Anthropic's statements are dated; the live catalog is the source of truth for a new request.

## Claude Fable 5.2 at a glance

| Field | What the catalog lists (2026-11-11) |
| --- | --- |
| Catalog id | `anthropic/claude-fable-5.2` |
| Short id (alias) | `claude-fable-5-2`, Anthropic's own model id |
| Input / output | text and image in, text out |
| Capability flags | chat, streaming, tool calling, vision |
| Endpoints | `POST /v1/messages` (Anthropic-native, recommended) and `POST /v1/chat/completions` |
| Not served on | `POST /v1/responses` |
| Plans | no Pro, Max or Ultra tier (2026-11-11 plan response) — wallet billing |
| Rates | per token, posted on the [model page](https://router.one/models/claude-fable-5-2); this guide prints no figures |

Router One serves Claude-family ids on `/v1/messages` and `/v1/chat/completions` only; the [API compatibility facts](https://router.one/facts/api-compatibility.md) state the endpoint rule for each model family.

## What Anthropic documents for Claude Fable 5.2

Anthropic's [release notes](https://platform.claude.com/docs/en/release-notes/overview) (November 9, 2026 entry) and its [Claude Fable 5.2 overview](https://platform.claude.com/docs/en/models/fable-5-2/overview), both checked 2026-11-11, describe the model on Anthropic's own API:

- **Release and id.** Anthropic launched Claude Fable 5.2 on November 9, 2026, with the model id `claude-fable-5-2`, and describes it as "built for demanding reasoning and long-horizon agentic work". Anthropic's [models overview](https://platform.claude.com/docs/en/models/overview) (checked 2026-11-11) now lists Claude Fable 5.1 and Claude Fable 5 among the legacy models that are still available.
- **Context and output.** A 1M-token context window and up to 128K output tokens, the limit Anthropic gives for its synchronous Messages API.
- **Thinking and effort.** Adaptive thinking is always on and cannot be turned off: per Anthropic's [migration guide](https://platform.claude.com/docs/en/models/fable-5-2/migration-guide) (checked 2026-11-11), `thinking: {"type": "disabled"}` and manual extended thinking (`budget_tokens`) return a 400 error, and so does forced tool use (a `tool_choice` of `any` or `tool`). Anthropic's [effort docs](https://platform.claude.com/docs/en/build-with-claude/effort) (checked 2026-11-11) list all five effort levels for the model, give `high` as the API default and suggest starting there.
- **Sampling parameters.** Per the same migration guide, omit `temperature`, `top_p` and `top_k`, as on Claude Fable 5.1; Anthropic's thinking docs list the values that return a 400 error.
- **Prefix check.** Claude Fable 5.2 enforces Anthropic's prefix check for every account, whatever its age: a request that edits the `system` prompt, the `tools` or an earlier message and then sends back a Claude Fable 5.2 thinking block from a later turn returns a 400 error (unless the request opts into Anthropic's beta drop_block setting). Anthropic's migration guide adds that nothing changes when Claude Code or the Claude Agent SDK builds the conversation history.
- **Thinking blocks.** Thinking blocks that Claude Fable 5.2 produces work only in the account that produced them, or in an account linked to it; per the migration guide, Anthropic's API drops such a block when another account sends it, and the request succeeds. Per Anthropic, Claude Opus 5.5 and Claude Fable 5.1 do not read Claude Fable 5.2's thinking blocks, so a conversation that moves from Claude Fable 5.2 to either of them continues without that reasoning.
- **Computer use.** The model takes computer use through the `computer_toolset_20260801` toolset; on Anthropic's API, a request that declares the earlier `computer_20251124` tool returns a 400 error.
- **Advisor tool.** With Anthropic's advisor tool, which is in beta, a Claude Fable 5.2 executor accepts only Claude Fable 5.2 or Claude Mythos 5.2 as its advisor, and any other advisor returns a 400 error.

The same release note announces Claude Mythos 5.2, which Anthropic offers only to organizations verified through its verification programs; the 2026-11-11 Router One catalog lists no Claude Mythos id. Those are Anthropic's statements about its own API; Router One is responsible for the catalog entry and for how the gateway routes and bills it. Run the request shapes your code depends on (thinking settings, tool definitions, edited conversation history) on your own prompts before you move a workload.

## Claude Fable 5.2, Claude Fable 5.1 or Claude Opus 5.5?

This section compares what Router One controls (the plan allowance a call draws, the endpoints, the posted rates), not output quality: run your own prompts on each candidate before you move a workload. As of the 2026-11-11 catalog and plan response:

| Model | Catalog id | Plan tier | Endpoints on Router One |
| --- | --- | --- | --- |
| Claude Fable 5.2 | `anthropic/claude-fable-5.2` | none (2026-11-11 plan response), wallet billing | `/v1/messages`, `/v1/chat/completions` |
| Claude Fable 5.1 | `anthropic/claude-fable-5.1` | none (2026-11-11 plan response), wallet billing | `/v1/messages`, `/v1/chat/completions` |
| Claude Opus 5.5 | `anthropic/claude-opus-5.5` | none (2026-11-11 plan response), wallet billing | `/v1/messages`, `/v1/chat/completions` |

The three share the endpoints and the wallet, so on Router One the difference you control is the posted rate: compare the [Claude Fable 5.2](https://router.one/models/claude-fable-5-2), [Claude Fable 5.1](https://router.one/models/claude-fable-5-1) and [Claude Opus 5.5](https://router.one/models/claude-opus-5-5) model pages, or open [Claude Fable 5.2 vs Claude Opus 5.5](https://router.one/models/compare/claude-fable-5-2-vs-claude-opus-5-5), which renders both spec sheets and live rates side by side. If you want Claude calls to draw plan quota instead, Claude Opus 5 (`anthropic/claude-opus-5`) is in the Premium models tier of Pro, Max and Ultra as of the 2026-11-11 plan response.

Anthropic's own advice, from its models overview (checked 2026-11-11): start with Claude Opus 5.5 for most workloads, and use Claude Fable 5.2 for demanding reasoning and long-horizon agentic work, or when your evals on Claude Opus 5.5 at higher effort still fall short. Anthropic's [Claude Fable 5.1 overview](https://platform.claude.com/docs/en/models/fable-5-1/overview) (checked 2026-11-11) marks Claude Fable 5.1 as legacy but still available, and suggests migrating to Claude Fable 5.2.

What changes when you move, per Anthropic's migration guide (checked 2026-11-11):

- **From Claude Fable 5.1.** Anthropic calls the move mostly drop-in; the prefix check, the thinking-block rule, the computer use toolset and the advisor pairing above are the four changes that can break code.
- **From Claude Opus 5.5.** Among the changes Anthropic lists, the default effort goes from `medium` to `high` on its API, so a request that omits effort runs one level higher: set it explicitly. Anthropic's API lets Claude Fable 5.2 read Claude Opus 5.5's thinking blocks, but Claude Opus 5.5 does not read Claude Fable 5.2's.

## How billing works

Claude Fable 5.2 bills per token at the input and output rates posted on its [model page](https://router.one/models/claude-fable-5-2). Router One posts its own rates for each model, with pricing up to 90% off official prices on select models, and this guide prints no per-token figures because they go stale. As of the 2026-11-11 plan response no plan tier lists Claude Fable 5.2, so every call bills the wallet at the posted rates, plan or no plan.

Per Anthropic's migration guide (checked 2026-11-11), adaptive thinking is always on for this model. Thinking output is billed at the model's posted output rate, with no separate reasoning line ([pricing facts](https://router.one/facts/pricing.md)), so effort is a cost lever as well as a quality one: measure per task before you set a default. Anthropic prices its own API separately; the rate Router One charges for the id is the one on its model page.

Dashboard → Logs shows each call's model, tokens, wallet charge or plan quota used, latency and status ([per-request observability](https://router.one/llm-observability)), so you can add up a task's wallet charges from the calls it made rather than estimate them from rates.

## Do subscription plans cover Claude Fable 5.2?

Not as of the 2026-11-11 plan response: no tier of Pro, Max or Ultra lists `claude-fable-5-2`, so Claude Fable 5.2 calls bill per token to the wallet at the posted rates, whether or not you hold a plan. The same plan response lists Claude Fable 5.1 and Claude Opus 5.5 in no tier either, and Claude Opus 5 in the Premium models tier of all three plans. Plan model lists change; the [pricing page](https://router.one/pricing) shows the live lists.

## Claude Code and the fable alias

Per Claude Code's [changelog](https://code.claude.com/docs/en/changelog) and [model configuration docs](https://code.claude.com/docs/en/model-config) (Claude Code 2.1.322, checked 2026-11-11), from 2.1.257 until 2.1.321 the `fable` alias requests `claude-fable-5-1`, and from 2.1.321 it requests `claude-fable-5-2`; the same docs say Claude Fable 5.2 needs Claude Code 2.1.321 or later. Claude Code treats a gateway set through `ANTHROPIC_BASE_URL` as the Claude API, per its [gateway protocol docs](https://code.claude.com/docs/en/llm-gateway-protocol) (checked 2026-11-11), so a session pointed at Router One sends those ids. As of the 2026-11-11 catalog, Router One lists both models (catalog ids `anthropic/claude-fable-5.1` and `anthropic/claude-fable-5.2`) and accepts both official ids; as of the 2026-11-11 plan response, no plan tier lists Claude Fable 5.2 or Claude Fable 5.1, so these calls bill per token to the wallet.

Three ways to run a session on Claude Fable 5.2:

- `/model fable`, or `claude --model fable`, on Claude Code 2.1.321 or later (`claude --version` shows which one you run).
- `/model claude-fable-5-2`, which selects the model by its id.
- `ANTHROPIC_MODEL=claude-fable-5-2` for the session's main model, in your shell profile — or in the `env` block of `~/.claude/settings.json` if the one-click install wrote your setup, since the install script sets only `ANTHROPIC_BASE_URL` and `ANTHROPIC_AUTH_TOKEN`, not a model:

```bash
export ANTHROPIC_BASE_URL=https://api.router.one
export ANTHROPIC_AUTH_TOKEN=sk-your-api-key
export ANTHROPIC_MODEL=claude-fable-5-2
```

If you set `ANTHROPIC_DEFAULT_FABLE_MODEL=claude-fable-5-1` following earlier advice, you can delete that line, and the alias follows Claude Code to Claude Fable 5.2; keep it if you want the alias to stay on Claude Fable 5.1. In Claude Code, use the official id or the alias: those are the ids Claude Code's model configuration docs (checked 2026-11-11) give for Claude Fable 5.2, and per its gateway protocol docs (checked 2026-11-11), Claude Code assumes a 200K context window for a model id it does not recognize. The [Claude Code setup guide](https://router.one/blog/claude-code-setup-guide) covers a fresh install.

## Send the first request

1. **Create a key.** Dashboard → API Keys → Create Key. Keys look like `sk-...`. For a trial, give the key a `maxSpend` cap. A key's spend cap (maxSpend) limits what that key can charge to your wallet: once its cumulative wallet charges reach the cap, its wallet-billed requests return HTTP 402. Requests covered by your plan's quota draw from the account's shared plan quota and do not count toward any key's cap. Your other keys keep working ([per-key cost tracking](https://router.one/llm-cost-tracking)).
2. **Keep a wallet balance.** Claude Fable 5.2 calls bill the wallet whether or not you hold a plan (2026-11-11 plan response); top up in Dashboard → Deposit.
3. **Pick the base URL for your client.** Anthropic-native SDKs and tools take `https://api.router.one` (the SDK's `base_url`, or `ANTHROPIC_BASE_URL`) and call `/v1/messages` under it; OpenAI-compatible SDKs take `https://api.router.one/v1`.
4. **Call the model.** The examples use `claude-fable-5-2`; the catalog id `anthropic/claude-fable-5.2` works in the same field.

Messages API with curl, streaming, with the effort level set to `high`, the API default and the level Anthropic's effort docs (checked 2026-11-11) suggest starting at:

```bash
curl https://api.router.one/v1/messages \
  -H "x-api-key: sk-your-api-key" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-fable-5-2",
    "max_tokens": 32000,
    "stream": true,
    "output_config": {"effort": "high"},
    "messages": [{"role": "user", "content": "Read this incident timeline and write a root-cause analysis with three follow-up actions."}]
  }'
```

The same call from the Anthropic Python SDK (`pip install -U anthropic`) — only `base_url` and the key change. A response can open with one or more `thinking` blocks, so read content by block type, not by position:

```python
import anthropic

client = anthropic.Anthropic(
    base_url="https://api.router.one",
    api_key="sk-your-api-key",
)

with client.messages.stream(
    model="claude-fable-5-2",
    max_tokens=32000,
    output_config={"effort": "high"},
    messages=[{"role": "user", "content": "Read this incident timeline and write a root-cause analysis with three follow-up actions."}],
) as stream:
    message = stream.get_final_message()

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

Chat Completions from the OpenAI Python SDK (`pip install -U openai`), for OpenAI-compatible code — here the base URL ends in `/v1`:

```python
from openai import OpenAI

client = OpenAI(
    base_url="https://api.router.one/v1",
    api_key="sk-your-api-key",
)

response = client.chat.completions.create(
    model="claude-fable-5-2",
    max_tokens=32000,
    messages=[
        {"role": "user", "content": "Read this incident timeline and write a root-cause analysis with three follow-up actions."},
    ],
)
print(response.choices[0].message.content)
print(response.choices[0].finish_reason, response.usage)
```

Rules that hold on both endpoints:

- **Set `max_tokens` on every request.** Thinking counts toward it: per Anthropic's migration guide (checked 2026-11-11), a small cap can stop after a `thinking` block and before any text, so leave room for both. On Router One's Chat Completions, `max_completion_tokens` is accepted as the same cap (if both are sent, `max_completion_tokens` wins).
- **Stream long work.** `stream: true` returns output as it is generated ([streaming guide](https://router.one/llm-streaming)).
- **Keep a healthy wallet balance.** Before a request runs, the gateway reserves an estimate against your balance; when the balance cannot cover it, the gateway lowers `max_tokens` to fit, which can cut a long answer short. If it cannot cover even the request's initial reservation, the gateway returns HTTP 402 before the model runs.
- **Leave out `temperature`, `top_p` and `top_k`.** Per Anthropic's migration guide (checked 2026-11-11), omit all three on Claude Fable 5.2; the [Claude sampling parameter errors page](https://router.one/claude-sampling-parameters-400) collects Anthropic's rules.
- **End `messages` with a user turn.** Per Anthropic's migration guide (checked 2026-11-11), prefilling the assistant message returns a 400 error; put format and tone rules in the system prompt instead.
- **Send files inline.** Router One serves no Files API: pass images as base64 content blocks.

**Which endpoint.** `/v1/messages` is the recommended path: it is Anthropic's own wire format, the one Claude Code sends, and the one Anthropic's thinking-block and effort rules above are written for, so use it for multi-turn tool loops. `/v1/chat/completions` serves the same id for clients that speak only the OpenAI format.

5. **Read the trace.** Dashboard → Logs shows each call's model, tokens, wallet charge or plan quota used, latency and status ([per-request observability](https://router.one/llm-observability)). Retryable upstream failures are absorbed by [automatic fallback](https://router.one/llm-fallback) across the candidate routes for the same model; a Claude Fable 5.2 request is never answered by a different model.

A request the gateway refuses before any model is called has no row in Dashboard → Logs; its status code and error message say why. A 401 means the key was not accepted, a 402 that the wallet balance or the key's `maxSpend` cannot cover the request's initial reservation, and a 404 that the path or the model id does not exist, so copy the id exactly as the catalog lists it. The [error codes page](https://router.one/llm-api-error-codes) covers the rest.

## Which clients can call it

Claude Code and the Anthropic SDKs call Claude Fable 5.2 on `/v1/messages`: point them at `https://api.router.one` with a Router One key ([CLI setup](https://router.one/docs/guides/cli-setup), [Claude Agent SDK setup](https://router.one/integrations/claude-agent-sdk)). Any client that speaks Chat Completions and takes a custom base URL can call it on `/v1/chat/completions`: the official OpenAI SDKs ([OpenAI SDK setup](https://router.one/integrations/openai-sdk)) and the chat apps and frameworks in the [integrations directory](https://router.one/integrations). Give such a client the base URL `https://api.router.one/v1` and enter `claude-fable-5-2` as the model.

Codex CLI cannot call Claude Fable 5.2. Codex CLI sends Responses API requests to `/v1/responses`, and Router One does not serve Claude-family ids there: such a request is refused with HTTP 400 before any model is called ([API compatibility facts](https://router.one/facts/api-compatibility.md)). Keep Codex CLI on ids served on `/v1/responses`, such as GPT.

## FAQ

**What is the model ID for Claude Fable 5.2 on Router One?**
The catalog id is anthropic/claude-fable-5.2, as of the 2026-11-11 catalog, and the gateway also accepts Anthropic's own id, claude-fable-5-2, for the same catalog entry. Use claude-fable-5-2 in Claude Code and in Anthropic SDK code. Either id bills per token to the wallet, since no plan tier lists the model as of the 2026-11-11 plan response.

**How much does the Claude Fable 5.2 API cost?**
On Router One, per token at the input and output rates posted on the Claude Fable 5.2 model page, with pricing up to 90% off official prices on select models; this guide prints no figures because they go stale. Every call bills the wallet, since no plan tier lists the model as of the 2026-11-11 plan response, and thinking output is billed at the posted output rate.

**Is Claude Fable 5.2 included in Router One subscription plans?**
Not as of the 2026-11-11 plan response: no Pro, Max or Ultra tier lists it, so its calls bill per token to the wallet, plan or no plan. Claude Opus 5 is in the Premium models tier of all three plans as of the same plan response. The pricing page lists each plan's models.

**How do I use Claude Fable 5.2 in Claude Code?**
On Claude Code 2.1.321 or later, run /model fable or launch with claude --model fable; /model claude-fable-5-2 and ANTHROPIC_MODEL=claude-fable-5-2 select it by id. Per Claude Code's changelog and model configuration docs (Claude Code 2.1.322, checked 2026-11-11), the fable alias requests claude-fable-5-1 from 2.1.257 until 2.1.321 and claude-fable-5-2 from 2.1.321. Router One lists both models as of the 2026-11-11 catalog and accepts both official ids.

**I pinned the fable alias to Claude Fable 5.1 earlier. Do I still need that line?**
No. As of the 2026-11-11 catalog Router One lists Claude Fable 5.2 and accepts its official id, so you can delete the line and let the fable alias follow Claude Code 2.1.321 or later to claude-fable-5-2. Keep the line if you want the alias to stay on Claude Fable 5.1; as of the 2026-11-11 plan response neither model is in a plan tier, so both bill the wallet.

**What are the context window and maximum output of Claude Fable 5.2?**
Per Anthropic's Claude Fable 5.2 overview (checked 2026-11-11), a 1M-token context window and up to 128K output tokens on Anthropic's synchronous Messages API. Thinking counts toward max_tokens, so set it explicitly with room for both and check stop_reason: max_tokens means the cap was reached. The Router One model page shows the live catalog entry.

**Can I turn thinking off or set temperature on Claude Fable 5.2?**
Not on Anthropic's own API. Per Anthropic's migration guide (checked 2026-11-11), adaptive thinking is always on, and a disabled thinking type or a manual budget returns a 400 error; the effort parameter sets the depth. Anthropic's guide also says to omit temperature, top_p and top_k.

**Can Codex CLI or the Responses API call Claude Fable 5.2?**
No. Router One serves Claude-family ids on /v1/messages and /v1/chat/completions; /v1/responses does not serve them, and a request there is refused with HTTP 400 before any model is called. Codex CLI speaks only the Responses wire format, so keep it on ids served there, such as GPT.

**Should I use Claude Fable 5.2 or Claude Opus 5.5?**
Anthropic's models overview (checked 2026-11-11) suggests starting with Claude Opus 5.5 for most workloads and using Claude Fable 5.2 for demanding reasoning and long-horizon agentic work, or when evals on Claude Opus 5.5 at higher effort still fall short. On Router One both are listed as of the 2026-11-11 catalog, both are served on the same two endpoints, and neither is in a plan tier as of the 2026-11-11 plan response, so the difference you control is the posted rate on each model page.

**How do I pay for Claude Fable 5.2 calls?**
From the wallet: as of the 2026-11-11 plan response, every Claude Fable 5.2 call bills per token there. Top up with a card or Alipay through one hosted checkout, or with USDT/USDC on six chains (Tron, BSC, Ethereum, Polygon, Base, Arbitrum). No US credit card required.

## Next steps

- Open the [Claude Fable 5.2 model page](https://router.one/models/claude-fable-5-2) for the live rates and endpoints.
- Put it beside [Claude Opus 5.5](https://router.one/models/compare/claude-fable-5-2-vs-claude-opus-5-5) on the comparison page.
- Check each plan's allowance and the models each tier lists on the [pricing page](https://router.one/pricing).
- Set up the terminal agent with the [Claude Code setup guide](https://router.one/blog/claude-code-setup-guide).
- Getting a 400 for temperature, top_p or top_k? See the [Claude sampling parameter errors page](https://router.one/claude-sampling-parameters-400).

## See also

- Canonical page: https://router.one/blog/claude-fable-5-2-api-guide
- Pricing and subscription plans: https://router.one/pricing
- LLM API Gateway and Routing: https://router.one/llm-api-gateway
- All blog posts: https://router.one/blog
- Models and per-model token rates: https://router.one/models (markdown: https://router.one/models.md)
- API docs (markdown): https://router.one/docs.md
- Company facts: https://router.one/facts/company.md