fix: update AWS Bedrock Anthropic model specs to latest official documentation

[nangsontay]

Related GitHub Issue

Closes: #

Roo Code Task Context (Optional)

N/A

Description

Updates AWS Bedrock Anthropic model specifications to match the latest official AWS documentation.

Key changes:

• Updated maxTokens for Claude Sonnet 4 (anthropic.claude-sonnet-4-20250514-v1:0) from 8,192 to 64,000
• Updated maxTokens for Claude Opus 4 (anthropic.claude-opus-4-20250514-v1:0) from 8,192 to 64,000
• Updated Claude Sonnet 4.6 and Claude Opus 4.6 to use native 1M context window (removed beta flag requirement and tiered pricing)
• Updated guessModelInfoFromId heuristics for Claude 4 family in bedrock provider
• Updated related test expectations

Files changed:

• packages/types/src/providers/bedrock.ts – Primary model definitions
• src/api/providers/bedrock.ts – Provider implementation (guessModelInfoFromId)
• src/api/providers/__tests__/bedrock.spec.ts – Test updates

Source: https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards-anthropic.html

Test Procedure

• All 87 existing bedrock tests pass (57 bedrock + 30 bedrock-reasoning/inference-profiles)
• No new tests needed — existing tests were updated to reflect corrected values

Pre-Submission Checklist

• Issue Linked: This PR is linked to an approved GitHub Issue (see "Related GitHub Issue" above).
• Scope: My changes are focused on the linked issue (one major feature/fix per PR).
• Self-Review: I have performed a thorough self-review of my code.
• Testing: New and/or updated tests have been added to cover my changes (if applicable).
• Documentation Impact: I have considered if my changes require documentation updates (see "Documentation Updates" section below).
• Contribution Guidelines: I have read and agree to the Contributor Guidelines.

Screenshots / Videos

N/A — no UI changes.

Documentation Updates

• No documentation updates are required.
• Yes, documentation updates are required.

Additional Notes

This is a data-only fix aligning the model specs with the latest AWS Bedrock documentation for Anthropic Claude models.

Interactively review PR in Roo Code Cloud

[Copilot]

Pull request overview

Updates AWS Bedrock Anthropic model metadata so the app’s model selection, token limits, and pricing calculations align with the latest Bedrock model-card documentation.

Changes:

• Increased maxTokens for Claude Sonnet 4 / Opus 4 family entries to 64,000.
• Marked Claude Sonnet 4.6 and Opus 4.6 as having a native 1M context window (removed 1M beta/tier behavior in type definitions and updated test expectations).
• Adjusted Bedrock provider heuristics for Claude 4 in guessModelInfoFromId.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
packages/types/src/providers/bedrock.ts	Updates Bedrock Anthropic model definitions (token limits, 1M context handling, tiers).
src/api/providers/bedrock.ts	Updates heuristic defaults for Claude 4 model IDs when a model isn’t in the static table.
src/api/providers/tests/bedrock.spec.ts	Updates expectations for Claude Sonnet 4.6 native 1M context behavior and pricing.

Comments suppressed due to low confidence (4)

src/api/providers/bedrock.ts:299

• guessModelInfoFromId patterns for Claude 4 won’t match the AWS Bedrock Anthropic model IDs used elsewhere in this repo (e.g. anthropic.claude-sonnet-4-20250514-v1:0 contains claude-sonnet-4, not claude-4). As a result, unknown Claude 4 Sonnet IDs will fall back to the default model info rather than a Claude 4-specific guess. Consider matching claude-sonnet-4 (and/or using a regex for claude-.*-4) in addition to claude-4.

		// Define a mapping for model ID patterns and their configurations
		const modelConfigMap: Record<string, Partial<ModelInfo>> = {
			"claude-4": {
				maxTokens: 64_000,
				contextWindow: 200_000,
				supportsImages: true,
				supportsPromptCache: true,
			},
			"claude-3-7": {

src/api/providers/bedrock.ts:316

• The Claude 4 Opus heuristic key claude-4-opus doesn’t match AWS Bedrock Anthropic Opus IDs (they are claude-opus-4-..., e.g. anthropic.claude-opus-4-20250514-v1:0). This makes the heuristic ineffective for unknown Opus 4 IDs and can cause incorrect pricing/capability defaults. Consider also matching claude-opus-4 (and claude-opus-4-6 if maxTokens differs).

			"claude-4-opus": {
				maxTokens: 64_000,
				contextWindow: 200_000,
				supportsImages: true,
				supportsPromptCache: true,
			},

packages/types/src/providers/bedrock.ts:39

• In this model entry the pricing comments still refer to “≤200K context”, but contextWindow is now set to a native 1,000,000. This is confusing/misleading documentation—either update the comments to reflect native 1M pricing semantics or remove the context qualifier.

		maxTokens: 64_000,
		contextWindow: 1_000_000, // Native 1M context window
		supportsImages: true,
		supportsPromptCache: true,
		supportsReasoningBudget: true,
		inputPrice: 3.0, // $3 per million input tokens (≤200K context)
		outputPrice: 15.0, // $15 per million output tokens (≤200K context)
		cacheWritesPrice: 3.75, // $3.75 per million tokens
		cacheReadsPrice: 0.3, // $0.30 per million tokens

packages/types/src/providers/bedrock.ts:145

• This Opus 4.6 entry sets contextWindow to native 1,000,000, but the pricing comments still say “≤200K context”. Please update/remove those qualifiers so the entry is self-consistent.

		maxTokens: 128_000,
		contextWindow: 1_000_000, // Native 1M context window
		supportsImages: true,
		supportsPromptCache: true,
		supportsReasoningBudget: true,
		inputPrice: 5.0, // $5 per million input tokens (≤200K context)
		outputPrice: 25.0, // $25 per million output tokens (≤200K context)
		cacheWritesPrice: 6.25, // $6.25 per million tokens
		cacheReadsPrice: 0.5, // $0.50 per million tokens

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

PR description still contains the placeholder Closes: # <!-- Replace with the issue number --> instead of an actual linked issue number. Please update the PR description to reference the approved issue so it can be tracked/auto-closed correctly.