[zoomote]

​Created by Roomote. Follow up by mentioning @roomote, in the web UI, or in Discord.

Related GitHub Issue

Closes: no tracked issue yet. A user reported the gap in Discord: Claude Haiku 5.5 did not appear in the Bedrock model dropdown. The user could type us.anthropic.claude-haiku-5-5 by hand, but a manual model ID is an unknown model, so prompt caching was off and costs rose. Maintainers can link an issue here.

Description

This PR adds anthropic.claude-haiku-5-5 to the AWS Bedrock model catalog in packages/types/src/providers/bedrock.ts. Users can now select Claude Haiku 5.5 from the model dropdown. The known-model entry carries prompt caching, so cache pricing applies again.

The entry mirrors the Claude Opus 5.5 entry:

• 128,000 max output tokens and a native 1M context window. No beta header is required, so the model does not join BEDROCK_1M_CONTEXT_MODEL_IDS.
• Prompt caching with 512-token cache checkpoints, 4 checkpoints per request, and the fields system, messages, and tools.
• Pricing per 1M tokens: input $0.10, output $0.50, cache writes $0.125, cache reads $0.01. Prompts over 100,000 tokens bill at higher long-context rates. The single price fields model the standard rates only.
• The AWS model card states that adaptive thinking is on by default and can be turned off. The model therefore joins BEDROCK_THINKING_DISABLE_MODEL_IDS, and the adaptive-thinking request contract applies to it: no temperature, effort-based thinking when reasoning is on. The card lists Global inference support, so the model joins BEDROCK_GLOBAL_INFERENCE_MODEL_IDS. The card lists the Standard tier only, so the model does not join BEDROCK_SERVICE_TIER_MODEL_IDS.

The AWS pricing page renders its tables in the browser, so the catalog prices come from Anthropic's pricing page, which lists the same Claude list prices billed on Bedrock. The same values match the shipped Opus 5.5 and Haiku 4.5 entries. Sources:

• https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-anthropic-claude-haiku-5-5.html
• https://platform.claude.com/docs/en/about-claude/pricing

Test Procedure

• Run pnpm --dir src exec vitest run api/providers/__tests__/bedrock.spec.ts api/providers/__tests__/bedrock-inference-profiles.spec.ts. bedrock.spec.ts covers the Haiku 5.5 catalog entry, cache flags and pricing, us. and global. prefix resolution, the thinking-disable path, and the adaptive-thinking guard. bedrock-inference-profiles.spec.ts is pre-existing prefix coverage. It does not test Haiku 5.5. All 162 tests pass.

• Run pnpm check-types. All 11 tasks pass.

• Run pnpm test. All 13 workspace tasks pass.

• Manual check: build the extension, open Zoo Code settings, pick Bedrock, and open the Model dropdown. anthropic.claude-haiku-5-5 appears between Haiku 4.5 and Opus 4.1, and selecting it shows prompt caching with the prices above.

Pre-Submission Checklist

• Issue Linked: No approved GitHub Issue exists yet. The Discord report above is the source. Maintainers can link an issue.
• Scope: The changes cover only the Bedrock Haiku 5.5 addition.
• Self-Review: The diff was reviewed line by line.
• Writing Style: The title and description use short sentences, plain words, and active voice.
• Testing: Tests cover the catalog entry, prefix resolution, thinking behavior, and the adaptive-thinking guard.
• Visual Snapshot: Not needed. The dropdown layout and theme tokens are unchanged; the change adds one catalog option. A dropdown screenshot ships in the paired docs PR.
• Documentation Impact: A paired docs PR updates docs/providers/bedrock.md. See Related PRs.
• Contribution Guidelines: Read and followed.

Visual Snapshots

No new rendered state needs a *.visual.tsx baseline. The dropdown component is unchanged. A screenshot of the dropdown with Claude Haiku 5.5 is in the paired docs PR.

Documentation Updates

Yes. The paired docs PR updates the Bedrock provider page. See Related PRs.

Related PRs

• Zoo-Code-Docs (Bedrock provider page, dropdown screenshots): [Docs] Document Claude Haiku 5.5 for AWS Bedrock Zoo-Code-Docs#69

Additional Notes

The Zoo-Code-Jetbrains plugin carries no Bedrock model list of its own. It consumes this extension through the deps/vscode submodule, so it inherits this change when that submodule is bumped.

Get in Touch

@roomote

[coderabbitai]

Important

Review skipped

Bot user detected.

To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

• Configuration used: Repository: Zoo-Code-Org/Zoo-Code/.coderabbit.yaml
• Review profile: ASSERTIVE
• Plan: Advanced
• Run ID: 57c4bfbd-6a00-4548-b6ac-aa4045e9f3b1

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

Review status

This PR was opened by an automated account. A human maintainer must verify the change intent, provenance, and validation before merging.

Current step: Required CI passed. Waiting for automated review of the latest commit.

If automated review does not start, a maintainer must restart it.

Review-state labels are managed by this workflow; do not edit them manually. community-approved is managed the same way — do not add or remove it manually. It signals a fresh community code approval for the current head as an advisory priority only; maintainer review is still required.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!