Kiro adds Claude Haiku 5.5 in AWS GovCloud (US), completing the Claude 5.5 family for the agentic IDE and CLI there

Amazon Web Services (Kiro); Anthropic model · Government cloud; developer agents · United States (AWS GovCloud US Regions)

AWS

Publisher: AWS What's New
Original publication: 2026-11-02
Source retrieved: 2026-11-03

Read original source

What happened

AWS announced on Nov 2 that Claude Haiku 5.5 is available in the Kiro IDE and CLI in the AWS GovCloud (US) Regions, so all three Claude 5.5 models are now in Kiro there. AWS positions Haiku 5.5 for high-volume, low-latency work such as summarising long context, classifying and routing requests, bulk text processing, small multi-file edits and parallel review subagents, with Opus 5.5 and Sonnet 5.5 doing harder planning. It has a 1M context window, a 0.06x credit multiplier and a 0.3x multiplier for requests over 100K tokens. The archive already holds Haiku 5.5 on Bedrock in GovCloud; this item is the Kiro developer-tool availability.

Why it matters

State and university development teams working inside GovCloud can now pick a cheaper, faster model tier for routine agentic coding tasks without leaving the boundary. Relevance is limited to organisations already using Kiro in GovCloud; no SLED customer or measured result is given.

Evidence and measured results

One-page vendor availability notice. No benchmark, accuracy, cost-per-task or customer data.

Limitations and uncertainty

Vendor notice only. Authorization of Kiro in GovCloud for a given workload remains the agency's responsibility. Relevance limited to existing Kiro GovCloud users.

Put this evidence to work

Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-11-03; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.

Sales

Role takeaway

Customer problem: GovCloud development teams need to control agent credit spend while keeping routine coding work inside the boundary.
Stakeholders: Application development manager
State or university CIO
CISO
Budget office
Discovery questions: Are you already using Kiro in GovCloud, and on which tasks?
How are agent credits budgeted today?
Which tasks would you accept a smaller model reviewing?
Value hypothesis: Routing bulk and review tasks to a cheaper tier may lower credit use for teams already on Kiro in GovCloud; savings are unmeasured.
Potential engagement: A two-to-four-week model-tier comparison on one team's backlog, ending in a tier policy.
Unsupported claims: Productivity or quality gains
Workload authorization for Kiro
Any specific cost saving

Pre-sales engineering

Role takeaway

Fit: Teams already running Kiro in GovCloud; not relevant to non-Kiro or commercial-only users.
Architecture and integration: Selected in Kiro IDE/CLI model picker; pairs with Opus/Sonnet 5.5 for planning in multi-agent workflows.
Prerequisites: Kiro in GovCloud enabled
Latest IDE/CLI version
Credit monitoring configured
Constraints: GovCloud US Regions only for this notice
Long requests over 100K tokens bill at 0.3x
No published benchmarks
Security: Confirm logging of model choice and that agency data handling rules for Kiro apply equally to the new tier.
Proof of value: Compare defect and rework rates against Sonnet 5.5 on the same tasks
Measure credit use per task
Verify admin control over model selection

Delivery

Role takeaway

Work: Add a model-tier policy and run a measured comparison within the existing Kiro rollout.
Dependencies: Kiro authorization in the agency boundary
Credit budget
Code review process
Ownership: Application development manager owns tier policy; CISO approves data-handling scope.
Skills and adoption: Developers need guidance on when to use each model tier and how to review subagent output.
Governance checkpoints: Policy approval before broad use
Review of comparison results
Re-check after AWS pricing or model changes
Acceptance criteria: Proposed: credit use per completed task tracked for all pilot tasks
Proposed: rework rate on Haiku tasks no higher than the baseline tier
Risks: Weaker review quality from smaller model
Cost surprises on long-context requests
Policy drift as models change

Implementation considerations

Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.

Architecture and integration

What must connect, and where does the AI sit in the workflow?

Model is selected in the Kiro IDE/CLI model picker after updating; work can be split so Haiku handles bulk or review subagent tasks while larger models plan.

Governance

Who approves, reviews and stays accountable for outcomes?

Agencies should record which model tier each agent task uses, because cheaper models used as reviewers change the assurance of code review.

Security and privacy

What data, permissions and controls need testing?

Remains inside GovCloud per the notice; the notice does not discuss data handling, logging or authorization status of Kiro itself.

Accessibility and workforce

Who is affected, and what skills or accommodations follow?

Shifts developer effort toward orchestrating and reviewing agent output; no accessibility findings.

Procurement

What should contracts, pricing and exit terms secure?

Credit multipliers (0.06x; 0.3x above 100K tokens) affect consumption budgeting; agencies should model spend by task type.

Operating model

Which teams own the service once it runs?

Supports a tiered model strategy for developer agents, which needs a documented policy on which tasks each tier may do.

Publication history

2026-11-02AWS · Issue 055 resources

Read preserved resource versions (JSON)

Stable resource ID: aws-kiro-claude-haiku-5-5-govcloud-20261102