feat(aws-bedrock): update model YAMLs [bot]

[models-bot]

Auto-generated by poc-agent for provider aws-bedrock.

---

Note

Medium Risk
Wide metadata changes affect routing, cost estimation, and allowed request params across many Bedrock models; incorrect limits or deprecation flags could mislead callers until validated against AWS docs.

Overview
Bulk refresh of aws-bedrock model catalog YAMLs to match current Bedrock docs and pricing.

Lifecycle: Both AI21 Jamba large/mini models move from active to deprecated with deprecationDate: "2026-05-26"; Jamba large token costs are normalized to scientific notation.

Params & capabilities: Many Nova chat profiles gain a documented reasoning_effort param (defaults and allowed values vary by model/region). APAC Nova Lite output caps drop from 10k → 5k tokens, aligned with max_tokens params. EU Nova 2 Lite cost entries add cache read and batch input/output pricing. thinking: true is added or adjusted on several Nova/DeepSeek profiles; US Nova Lite drops thinking and strips reasoning_effort via removeParams. Global Claude Opus 4.6 adds structured_output in features.

Anthropic pricing/limits: Claude Haiku/Opus/Sonnet entries across regions add cache_creation_input_token_cost_per_hour. tool_use_system_prompt_tokens increases from 346 → 496/497 on multiple Claude 4.5/4.6 profiles.

Embeddings & non-chat: Embedding, image, and several instruct models add reasoning_effort to removeParams so clients do not send unsupported params. Cohere embed English v3 adds image input modality; EU Cohere embed v4 adds missing regions.

Other corrections: Token/context limits and regional costs are tweaked (e.g. Mistral Ministral 3B 8192 → 8000, Writer Palmyra X5 context ~1M → 128k, DeepSeek R1 max output 8000 → 8192, Voxtral context 32768 → 32000, Mistral Large eu-west-2 pricing). Minor doc sources updates and param key cleanup (e.g. n unquoted).

^{Reviewed by Cursor Bugbot for commit 1e90bd1. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

/test-models

[harshiv-26]

/test-models

[github-actions]

/test-models

[github-actions]

/test-models

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

There are 2 total unresolved issues (including 1 from previous review).

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 1e90bd1. Configure here.}