feat(aws-bedrock/eu.anthropic.claude-sonnet-5-5): add new models [bot] - #4097
models-bot[bot] wants to merge 2 commits into
Conversation
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes using default effort and found 2 potential issues.
❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.
Reviewed by Cursor Bugbot for commit 6f4af14. Configure here.
| - high | ||
| - xhigh | ||
| - max | ||
| type: string |
There was a problem hiding this comment.
Inherited max_tokens cap too low
High Severity
This fuller config sets max_output_tokens to 128000 but does not override inherited max_tokens maxValue. Provider default.yaml caps max_tokens at 2048, so the gateway clamps or rejects outputs far below the model's 128000 limit.
Triggered by learned rule: Override max_tokens maxValue to match limits
Reviewed by Cursor Bugbot for commit 6f4af14. Configure here.
| cache_read_input_token_cost: 2.2e-7 | ||
| input_cost_per_token: 0.0000022 | ||
| output_cost_per_token: 0.000011 | ||
| region: eu-west-3 |
There was a problem hiding this comment.
Fuller costs omit batch rates
Medium Severity
This fuller Claude cost table has token and cache rates but omits input_cost_per_token_batches and output_cost_per_token_batches. Batch inference estimates miss the Claude Batch 50% discount and price jobs at on-demand rates.
Triggered by learned rule: AWS Bedrock per-id metadata is not a sibling copy
Reviewed by Cursor Bugbot for commit 6f4af14. Configure here.


Auto-generated by model-addition-agent for
aws-bedrock/eu.anthropic.claude-sonnet-5-5.Note
Low Risk
Adds static model metadata only; no runtime or auth logic changes.
Overview
Adds a new provider catalog file for
eu.anthropic.claude-sonnet-5-5on AWS Bedrock so routing, pricing, and capability metadata can target the EU inference profile separately from existingeu.anthropic.claude-sonnet-5andglobal.anthropic.claude-sonnet-5-5entries.The definition marks the model active and serverless, with per-token costs (including prompt-cache fields) for eight EU regions, chat mode, text/image input, thinking enabled, a reasoning_effort parameter (default
high), and standard Claude tool/caching features. Limits are set to a 1M context window and 128k max output tokens.Reviewed by Cursor Bugbot for commit 6f4af14. Bugbot is set up for automated code reviews on this repo. Configure here.