feat: support claude-sonnet-4-5-20250929

[prnake]

Summary by CodeRabbit

• New Features

  • Added support for Claude Sonnet 4.5 (20250929) and its Thinking variant.
  • Enabled on multiple providers, including AWS and Vertex AI.
  • Cross-region availability enabled in US, EU, and AP regions.

• Chores

  • Updated default cache and create-cache ratios for the new models.
  • Updated model routing ratios to include the new models.

[coderabbitai]

Walkthrough

Adds new Claude model identifiers (claude-sonnet-4-5-20250929 and its -thinking variant) across providers and settings: updates AWS and Vertex model mappings, extends AWS cross-region allowances, updates Claude channel model list, and adjusts cache/model ratio maps.

Changes

Cohort / File(s)	Summary
AWS model mappings and regions relay/channel/aws/constants.go	Mapped claude-sonnet-4-5-20250929 to anthropic.claude-sonnet-4-5-20250929-v1:0; allowed cross-region use in us, ap, eu.
Claude channel model list relay/channel/claude/constants.go	Added claude-sonnet-4-5-20250929 and claude-sonnet-4-5-20250929-thinking to ModelList.
Vertex model mapping relay/channel/vertex/adaptor.go	Added mapping: claude-sonnet-4-5-20250929 → claude-sonnet-4-5@20250929.
Cache ratios for new models setting/ratio_setting/cache_ratio.go	Added default cache ratios and create-cache ratios for the two new models, matching existing peers.
Model ratio update setting/ratio_setting/model_ratio.go	Added claude-sonnet-4-5-20250929 with ratio 1.5 to defaultModelRatio.

Sequence Diagram(s)

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Possibly related PRs

• feat: add support for claude-opus-4-1 model and update ratios #1511 — Parallel additions for new Claude models across AWS/Vertex mappings, Claude list, and ratio maps.
• feat: 新增支持目前已发布的amazon nova model #1882 — Similar AWS constants updates for model ID mapping and cross-region entries.

Suggested reviewers

• seefs001

Poem

New sonnets hop across the sky,
Mapped in clouds where regions lie.
Vertex hums, AWS replies—
Ratios set, the caches rise.
A rabbit stamps a tiny paw:
“New models in, no extra flaw.” 🐇✨

Pre-merge checks and finishing touches

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title Check	✅ Passed	The title clearly and concisely summarizes the primary change of the pull request, which is adding support for the new “claude-sonnet-4-5-20250929” model, and uses a conventional commit prefix to indicate it’s a feature addition.
Docstring Coverage	✅ Passed	No functions found in the changes. Docstring coverage check skipped.

✨ Finishing touches

• 📝 Generate Docstrings

🧪 Generate unit tests

• Create PR with unit tests
• Post copyable unit tests in a comment

---

📜 Recent review details

Configuration used: CodeRabbit UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 4bb4b64 and d7db30a.

📒 Files selected for processing (5)

• relay/channel/aws/constants.go (2 hunks)
• relay/channel/claude/constants.go (1 hunks)
• relay/channel/vertex/adaptor.go (1 hunks)
• setting/ratio_setting/cache_ratio.go (2 hunks)
• setting/ratio_setting/model_ratio.go (1 hunks)

🧰 Additional context used

🧠 Learnings (1)

📚 Learning: 2025-08-05T17:14:17.246Z

Learnt from: neotf
PR: QuantumNous/new-api#1511
File: setting/ratio_setting/model_ratio.go:118-123
Timestamp: 2025-08-05T17:14:17.246Z
Learning: Claude models handle "-thinking" variants differently from Gemini models. For Claude models, only the base model (without "-thinking") gets an entry in defaultModelRatio map. The "-thinking" variants rely on the Claude relay handler stripping the suffix using strings.TrimSuffix(textRequest.Model, "-thinking") before looking up the ratio, so they automatically use the base model's ratio.

Applied to files:

• setting/ratio_setting/model_ratio.go

🔇 Additional comments (7)

setting/ratio_setting/model_ratio.go (1)

144-144: LGTM! Correctly follows Claude model ratio pattern.

The addition of the base model entry without a separate "-thinking" variant is correct. Based on learnings, Claude's relay handler strips the "-thinking" suffix before lookup, so the thinking variant will automatically use this base ratio.

relay/channel/claude/constants.go (1)

22-23: LGTM! Model identifiers added correctly.

Both the base model and thinking variant are properly added to the ModelList, consistent with the existing pattern for Claude Sonnet 4 models.

relay/channel/vertex/adaptor.go (1)

40-40: LGTM! Vertex AI model mapping added correctly.

The mapping follows the established pattern, converting the model identifier to Vertex AI's naming convention (replacing date hyphens with @).

relay/channel/aws/constants.go (2)

19-19: LGTM! AWS model ID mapping added correctly.

The mapping follows the established naming pattern for Anthropic Claude models on AWS Bedrock.

---

73-77: LGTM! Cross-region configuration is appropriate.

The regional availability (us, ap, eu) matches other Claude Sonnet 4 models, providing consistent multi-region support.

setting/ratio_setting/cache_ratio.go (2)

55-56: LGTM! Cache ratio entries added correctly.

Both the base model and thinking variant entries are properly added with consistent ratios (0.1), matching the pattern of other Claude Sonnet 4 models.

---

74-75: LGTM! Create cache ratio entries added correctly.

Both entries use the standard 1.25 ratio, consistent with other Claude models in the create cache configuration.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

🧪 Early access (Sonnet 4.5): enabled

We are currently testing the Sonnet 4.5 model, which is expected to improve code review quality. However, this model may lead to increased noise levels in the review comments. Please disable the early access features if the noise level causes any inconvenience.

Note:

• Public repositories are always opted into early access features.
• You can enable or disable early access features from the CodeRabbit UI or by updating the CodeRabbit configuration file.

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[neotf]

LGTM