feat: Support Custom Metadata Labels on Vertex AI

[yu-iskw]

Fixes # (issue) #8205

This is already supported in langchain-google. So, it would be great to support it in langchainjs as well.

https://github.com/langchain-ai/langchain-google/blob/24ce2fd6aecf7e06c4f524a60e0bb1671e6516c2/libs/vertexai/langchain_google_vertexai/chat_models.py#L1184-L1186

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
langchainjs-docs	✅ Ready (Inspect)	Visit Preview		May 27, 2025 4:19am

1 Skipped Deployment

Name	Status	Preview	Comments	Updated (UTC)
langchainjs-api-refs	⬜️ Ignored (Inspect)			May 27, 2025 4:19am

[benjamincburns]

@afirstenberg any thoughts on this?

[afirstenberg]

I think this is ok, but would like to see a test to confirm it doesn't show up if not specified.

[afirstenberg]

Thanks for the PR! I hadn't seen this in the Vertex docs before, but it does look useful.

Overall, this looks good to me. One comment about making sure the labels aren't included if not set.

We should run a test to see if it works on AI Studio, tho it doesn't look it from the discovery doc. I don't think we need to explicitly handle that case, however.

I'd rather see the tests integrated into the current test suite (particularly in google-common:chat_models.test.ts using one of the mocked responses and google-vertex/chat_models.int.test.ts)

[yu-iskw]

@afirstenberg Thank you for the feedback. I have updated the code. Can you review it again please?

According to my research, the feature of custom metadata labels is currently supported only in Vertex AI. Gemini API doesn't have the feature at the moment. We can also implement the feature only in langchain-google-vertexai and langchain-google-vertexai-web. In that case, we need to implement the property in each. So, I implemented the property in langchain-google-common.