Can I use Claude Fable 5 on AWS Bedrock right now?

Fable 5 was restored globally on July 1, 2026, and AWS is re-enabling it on Bedrock region by region. Depending on your region and account you may still see the gated flow from the late-June window: US-entity identity verification on the AWS account plus a use case approved by Anthropic. If calls fail, check Model access in your exact region — or use the direct Claude Platform API, which is live worldwide.

What does Anthropic ask for in the use case?

A short, specific description: who you are, what you will build, expected token volume, your data handling, and the region. Keep it concrete and low-risk. Vague requests and anything resembling a public reseller or proxy are the most common rejection reasons.

How long does Bedrock model access approval take?

Anthropic reviews accounts individually, so it varies from hours to days. Submit early, keep the use case specific, and use a cheaper model on Bedrock (Opus 4.8) in the meantime so you are not blocked.

Which region should I enable Claude Fable 5 in?

us-east-1 is the most reliable starting point. Model access is per-region — being approved does not auto-enable the model everywhere, so enable it under Bedrock → Model access in the exact region you call from.

Back to blog

How to Get Claude Fable 5 Approved on AWS Bedrock (2026)

Fable 5 is restored globally and Amazon Bedrock is re-enabling it region by region. The approval process, a copy-paste use-case template, and how to call the model once your account is in.

Jun 25, 2026

Fable5 Editorial

How to Get Claude Fable 5 Approved on AWS Bedrock (2026)

Update — July 1, 2026: Fable 5 is restored worldwide — the US export-control directive was withdrawn on June 30, and the model is live again on the Claude Platform API, Claude.ai, Claude Code, and Cowork. AWS Bedrock (along with Google Cloud and Microsoft Foundry) is being re-enabled "as soon as possible", region by region. Through July 7, Pro/Max/Team plans include Fable 5 in up to 50% of the weekly usage cap; after that it moves to usage credits. Live status: /is-fable-5-back · Full story: Claude Fable 5 is back.

Claude Fable 5 is back and it has reappeared in the Amazon Bedrock model catalog. But seeing it in the catalog is not the same as being able to call it. During the late-June window, access was gated: AWS wanted identity verification, and Anthropic approved use cases individually. Since the July 1 global restoration, Bedrock re-enablement is rolling out region by region — and the same approval mechanics can still apply while it does. This is the practitioner's walkthrough — what the gate actually is, how to get through it fast, and how to call the model once you're approved.

Why you can see Fable 5 but can't call it

In late June — roughly two weeks into the suspension — Fable 5 reappeared in the Bedrock catalog under tighter conditions, ahead of the full July 1 restoration. On Bedrock that showed up as a two-part gate:

AWS identity verification. The account (or organization) needs US-entity identity verification on file. This is AWS's side of the control.
Anthropic use-case review. Even with a verified account, Anthropic reviews a short use case you submit and approves accounts individually. Only approved accounts can invoke claude-fable-5.

Until both clear, the model card renders, "Open in playground" may even look active, but invocation returns an access error. That gap was the single biggest source of confusion during the rollout — and can still bite while AWS finishes re-enabling regions after the July 1 restoration.

Step 1 — Verify the account

In the AWS console, confirm your account or organization has completed the required identity verification for a US entity. Gated Anthropic frontier models are restricted to verified accounts, so this has to be done before a model-access request will go anywhere.

Step 2 — Request model access in Bedrock

Go to Bedrock → Model access, find Anthropic — Claude Fable 5, and request access. This is where you attach the use case. Treat the form as the thing a human will actually read.

Step 3 — Write a use case that gets approved

The fastest approvals share a pattern: a real organization, a bounded workload, honest volume, and clear data handling. Copy this, fill the brackets, and keep it to a short paragraph:

Company / project: <your org or product name>
Primary use case: <e.g. internal coding assistant for a 12-person dev team>
Why Fable 5: long-horizon agentic coding and large-refactor reasoning that
  smaller models cannot finish in one pass.
Expected monthly volume: <e.g. ~40M input / 2M output tokens>
Data handling: prompts contain our own source code only; no end-user PII.
Safety: requests routed through our backend; no public, unmoderated proxy.
Region: <e.g. us-east-1>

What reviewers are screening for is risk. Name a real entity, describe a specific job, state that you are not running an open proxy, and be honest about volume and data. The most common rejections are vague one-liners ("general AI tasks") and anything that reads like reselling access.

Step 4 — Enable the model after approval

Once approved, the model still has to be enabled per region. Open Bedrock → Model access in your target region (start with us-east-1) and toggle Claude Fable 5 on. Now the model ID is callable and the playground works for real.

Step 5 — Call it

From the AWS CLI or your SDK:

aws bedrock-runtime invoke-model \
  --model-id anthropic.claude-fable-5-v1:0 \
  --region us-east-1 \
  --body '{"anthropic_version":"bedrock-2023-05-31","max_tokens":1024,"messages":[{"role":"user","content":"Hello, Fable 5"}]}' \
  out.json

To drive Claude Code through Bedrock instead of the direct API:

export CLAUDE_CODE_USE_BEDROCK=1
export ANTHROPIC_MODEL=anthropic.claude-fable-5-v1:0

If a call returns an access error after approval, the usual cause is region: the model is enabled in one region but you're calling another. Re-check Model access in the exact region.

What it costs and how to plan

Bedrock pricing mirrors the direct API — $10 per million input tokens and $50 per million output, the same pricing math as elsewhere. Agent workloads are cache-heavy, so real costs land below the 2x sticker versus Opus 4.8. While you wait for approval or your region's re-enablement, Opus 4.8 on Bedrock is a near drop-in — change the model ID and keep everything else. Or skip the wait entirely: the direct Claude Platform API has been live worldwide since July 1. More options in our alternatives guide.

The smartest move during any wait is to get your prompt ready. Build and test a Fable 5 system prompt now; the same prompt works on Bedrock, the direct API, and Claude Code, so you ship the moment access lands.

FAQ

Is Claude Fable 5 available on AWS Bedrock? Yes — Fable 5 was restored globally on July 1, 2026, and AWS is re-enabling it on Bedrock region by region. Depending on your region and account, identity verification plus an Anthropic-approved use case may still be required to invoke it.

Why is approval required? Fable 5 is a frontier Mythos-class model that returned under tighter access conditions in the late-June window. As the post-restoration rollout completes, verification requirements may relax — check the console for your account's current state.

What if I'm rejected? Resubmit with a specific organization, a bounded workload, clear data handling, and realistic volume. Use Opus 4.8 on Bedrock in the meantime — or the direct Claude Platform API, which is live worldwide.

Independent analysis, last updated July 2, 2026. This rollout is moving fast — confirm against the AWS and Anthropic consoles, and our live Is Fable 5 back? tracker. Not affiliated with Anthropic or AWS.