Claude Code 2.1.94 ships Bedrock (Mantle…

ANTHROPIC PUB_DATE: 2026.04.08

CLAUDE CODE 2.1.94 SHIPS BEDROCK (MANTLE) SUPPORT; 2.1.96 HOTFIXES BEDROCK AUTH REGRESSION

Anthropic’s Claude Code added Amazon Bedrock (Mantle) support in 2.1.94 and fixed a Bedrock auth regression in 2.1.96 amid reliability debate. The [v2.1.94 rel...

Anthropic’s Claude Code added Amazon Bedrock (Mantle) support in 2.1.94 and fixed a Bedrock auth regression in 2.1.96 amid reliability debate.

The v2.1.94 release adds Amazon Bedrock powered by Mantle (enable with CLAUDE_CODE_USE_MANTLE=1) and bumps the default effort level from medium to high for API key, Bedrock/Vertex/Foundry, Team, and Enterprise users (tune with /effort). It also includes numerous stability fixes and small UX tweaks.

A follow-up v2.1.96 hotfix resolves Bedrock requests failing with 403 "Authorization header is missing" when using AWS_BEARER_TOKEN_BEDROCK or CLAUDE_CODE_SKIP_BEDROCK_AUTH, a regression introduced in 2.1.94.

Meanwhile, some enterprise developers question Claude Code’s reliability for complex, multi-file work per InfoWorld and TechRadar; evaluate in your workflows before broad rollout. For a broader context on recent changes, see this Q1 round-up on AI Maker.

[ WHY_IT_MATTERS ]

01.

Bedrock (Mantle) support simplifies enterprise adoption routes and centralizes access controls for teams already on AWS.

02.

The default effort shift to high can change behavior, latency, and token usage; plan for cost and CI/CD stability.

[ WHAT_TO_TEST ]

terminal
In staging, enable CLAUDE_CODE_USE_MANTLE=1 and verify Bedrock calls succeed on 2.1.96 with AWS_BEARER_TOKEN_BEDROCK; measure latency, throughput, and token costs.
terminal
Run the same repo tasks with /effort=medium vs /effort=high to quantify differences in edit scope, token usage, and runtime.

[ BROWNFIELD_PERSPECTIVE ]

Legacy codebase integration strategies...

01.
If you pinned 2.1.94 and saw Bedrock 403s, upgrade to 2.1.96; review any use of CLAUDE_CODE_SKIP_BEDROCK_AUTH and related auth flows.
02.
Audit pipelines for behavior changes from the new default effort level; set explicit /effort or config to avoid surprise diffs and spend.

[ GREENFIELD_PERSPECTIVE ]

Fresh architecture paradigms...

01.
If you’re standardizing on AWS, start with Bedrock via Mantle for simpler governance; document required env vars in your templates.
02.
Define organizational defaults for effort and codegen policies up front to keep edits focused and predictable.

arrow_back

PREVIOUS_DATA_LOG

Grounding, Sandboxing, and Streaming: Making AI Agents Production-Ready for Backend Teams

Initialize_Return_to_Core

LINK_STATUS: 127.0.0.1 (SECURE)

NEXT_DATA_LOG

Cursor 3 goes agent-first with a standalone workspace; pilot before you roll it out

arrow_forward