How to Access Claude Fable 5.1 via API: Model ID, Keys, and Rate Limits

Quick Answer

Claude Fable 5.1's API model ID is claude-fable-5-1. It speaks the standard Anthropic Messages API — same SDK, new model string — but ships with three behavior changes: forced tool calls now return 400 errors, thinking blocks can't cross model generations, and editing prior conversation turns is restricted. Pricing is $10/$50 per million tokens with cache reads at $0.25. Get a key from console.anthropic.com, or access it pay-as-you-go through TeamoRouter with Alipay/WeChat and no international card.

The Model ID

One clarification the search results keep muddying: Fable 5.1 is not subscription-only. It's on Max/Team/Enterprise subscription seats, but the API is open to all paid developers — "fable api only" claims usually refer to the fact that Mythos 5.1 is gated, not Fable.

First Call

import anthropic

client = anthropic.Anthropic(api_key="sk-ant-your-key")

message = client.messages.create(
    model="claude-fable-5-1",
    max_tokens=2048,
    messages=[{"role": "user", "content": "Explain cache-aside vs write-through in two paragraphs."}],
)
print(message.content[0].text)

Through a gateway, only the credentials change:

client = anthropic.Anthropic(
    base_url="https://api.teamorouter.com/anthropic",
    api_key="tr-your-key",
)
# everything else identical

Node.js (@anthropic-ai/sdk), REST, and the OpenAI-compatibility layers all work as with previous Claude models — no SDK upgrade is required for 5.1.

The Three API Behavior Changes (vs Fable 5)

These are the things that will actually break code, per Anthropic's migration guide:

1. Forced tool calls return 400 errors.

# Fable 5 — worked; Fable 5.1 — 400 error
tool_choice = {"type": "any"}
tool_choice = {"type": "tool", "name": "get_weather"}

# Fable 5.1 — use auto + prompt steering
tool_choice = {"type": "auto"}

2. Thinking blocks don't cross generations. Fable 5.1's thinking blocks can't be passed back to older models. Strip them when routing history to non-5.1 models.

3. Editing conversation history is restricted. Rewriting or patching prior turns mid-conversation fails. Append corrective turns or fork from the last clean state instead.

Full fixes and a migration checklist: Fable 5.1 migration guide.

Using the Cache (Where the Savings Are)

5.1's headline API economics: cache reads at $0.25/M vs $10/M fresh input. Mark your stable prefix as cacheable:

message = client.messages.create(
    model="claude-fable-5-1",
    max_tokens=2048,
    system=[
        {
            "type": "text",
            "text": LONG_SYSTEM_PROMPT,
            "cache_control": {"type": "ephemeral"},
        }
    ],
    messages=[{"role": "user", "content": "..."}],
)
# check response.usage.cache_read_input_tokens to verify hits

Agent loops with 50K+ cached prefixes are where the 75% cut turns into real money — see the worked tables in Fable 5.1 pricing.

Rate Limits and Access Tiers

Anthropic tiered rate limits by usage history and spend tier; 5.1 inherits the standard tier structure (usage limits per tier are published in the console, and change — check your org's limits page rather than trusting blog numbers, including this one). Practical notes:

• New API accounts start in the lowest tier with tight per-minute limits — expect to ramp spend to unlock headroom.
• Subscription seats (Max / premium Team/Enterprise) get Fable 5.1 with up to half the weekly usage limit allocatable to it — good for interactive use, not for programmatic volume.
• Gateways abstract this: TeamoRouter pools upstream capacity, so a fresh account gets practical throughput on day one without waiting for tier upgrades.

Error Patterns to Expect

FAQ

What is the Claude Fable 5.1 API model ID?

claude-fable-5-1. The same string works through gateways like TeamoRouter; the gateway handles upstream routing.

Is there a Fable 5.1 free tier?

No permanent free API tier. Fable 5.1 is included in Max and premium Team/Enterprise subscription seats (up to half your weekly limit). For API access, pay-as-you-go through Anthropic or a gateway is the way in.

How much does the Fable 5.1 API cost?

$10 / $50 per million input / output tokens; cache reads $0.25/M. Worked cost examples: Fable 5.1 pricing.

Why do my forced tool calls fail with a 400?

Fable 5.1 removed forced tool use (tool_choice type any or tool) because it skips extended thinking. Switch to {"type": "auto"} and steer with prompts, or make the deterministic call in your own code.

Can I use Fable 5.1 from China?

Yes — through a gateway. TeamoRouter offers pay-as-you-go access with Alipay/WeChat billing, no international card or VPN required.

Does Fable 5.1 work with the OpenAI-compatible SDK format?

Through gateways that expose an OpenAI-compatible endpoint, yes. Direct Anthropic access uses the native Messages API and Anthropic SDKs.

Summary

Fable 5.1's API story: same SDK, new ID (claude-fable-5-1), three behavior changes to fix, and a cache-read price that rewards good prompt structure. Sign up for TeamoRouter to skip the tier ramp and call claude-fable-5-1 on day one.

Get Started

TeamoRouter — one key, one base_url, Fable 5.1 and every other flagship on pay-as-you-go.