Updated July 2, 2026
Claude Fable 5 on AWS Bedrock
With the export controls lifted, Anthropic is re-enabling Fable 5 on AWS Bedrock following the July 1, 2026 global restoration. Here is the current access picture and how to get your account ready.
Quick version: Fable 5 returned globally on July 1, 2026, and Anthropic says AWS, Google Cloud, and Microsoft Foundry access is being re-enabled as quickly as possible. If your Bedrock account was gated during the suspension (US identity verification + use-case review), those approvals carry over.
Current access status
After the June 12 – July 1 suspension, Claude Fable 5 is being restored on AWS Bedrock as part of the global redeployment. During the gated late-June window, AWS required US-entity identity verification plus an Anthropic-reviewed use case — if you completed that, you're set; if not, request model access in the Bedrock console (steps below) while the open rollout completes.
Status as of July 2, 2026 — an evolving rollout; verify against the AWS and Anthropic consoles, which override this page when they differ.
How to get access, step by step
- 1
Complete US identity verification
In the AWS console, ensure your account (or organization) has the required US-entity identity verification on file. Anthropic's gated frontier models are restricted to verified accounts.
- 2
Write a simple use case
Prepare a short, concrete description of how you will use Fable 5 — who you are, the workload, expected volume, and your data-handling. Use the template below.
- 3
Submit for Anthropic review
Submit the use case through the model-access request in Bedrock (Model access → request access for Anthropic Claude Fable 5). Anthropic reviews submissions and approves accounts individually.
- 4
Enable the model once approved
After approval, enable Claude Fable 5 under Bedrock → Model access for your region. The 'Open in playground' option becomes active and the model ID is callable.
- 5
Call it from the SDK or Claude Code
Invoke the Bedrock model ID from your app, the AWS CLI, or point Claude Code at Bedrock. See the setup snippet below.
Use-case request template
Approvals go faster when the use case is specific and low-risk. Copy this, fill the brackets, and keep it to a short paragraph — vague or proxy-reseller-style requests are the most common rejection reason.
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 complete 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>Tip: name a real organization and a bounded workload, state that prompts contain no end-user PII, and confirm you are not building a public unmoderated proxy. Honesty about volume and data handling is what reviewers look for.
Calling Fable 5 on Bedrock
Once your account is approved and the model is enabled in your region, invoke it like any other Bedrock model:
# After your account is approved, call the model on Bedrock:
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
# Or point Claude Code at Bedrock:
export CLAUDE_CODE_USE_BEDROCK=1
export ANTHROPIC_MODEL=anthropic.claude-fable-5-v1:0Availability rolls out by region — us-east-1 is the safest first choice. If invocation returns an access error after approval, confirm the model is enabled under Model access in that exact region.
Get your prompts ready before approval lands
Approval can take time. Build and test your Fable 5 system prompt now so you can ship the moment access opens — the same prompt works on Bedrock, the API, and Claude Code.