# What changed in @langchain/google-common

> 25 changes in @langchain/google-common 1 to 2 from its official release notes, upgrade guides and advisories: 1 security, 5 new APIs and options. Each one cited. Sources checked 2026-10-19. HTML: https://unstale.sh/npm/@langchain/google-common/

To see which of these apply to a repository, run `npx @unstale/cli init` in it.

## @langchain/google-common 2

### New APIs and options

- Adds the MINIMAL thinking level for Gemini 3 Flash models. (2.1.27) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.27)
- Structured output now supports standard schemas, allowing developers to use them with Google integrations. (2.1.24) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.24)
- Aborted invoke() calls during streaming throw ModelAbortError with accumulated partialOutput when using streaming callback handlers. (2.1.14) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.14)
- Gemini models now support the thinkingLevel parameter. (2.1.4) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.4)
- withStructuredOutput now supports the jsonSchema method, allowing structured output to be defined with JSON Schema. (2.1.0) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.0)

### Behavior and default changes

- Adds native Gemini event conversion for streamEvents. (2.2.0) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.2.0)
- Chat model and LLM invocation parameters are now added to traced metadata. (2.1.28) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.28)
- Packages now stamp their version into runnable metadata at construction time, making version information available in LangSmith trace metadata. (2.1.21) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.21)
- Model profiles were updated, which may affect how model capabilities are represented. (2.1.19) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.19)

### Fixes and other changes

- Fixes the Google tool choice modes sent to Vertex AI. (2.3.1) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.3.1)
- Fixes endpoint routing and parameter validation for the eu multi-region location. (2.3.0) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.3.0)
- Automatically sets this flag when a Gemini request mixes server-side built-in tools with function declarations, avoiding an HTTP 400 response. (2.2.0) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.2.0)
- Guards against plain object ToolMessage content to prevent a crash. (2.2.0) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.2.0)
- Fixed missing stream usage metadata on Gemini. (2.1.31) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.31)
- Prevents a 400 error when an AIMessage has empty text content. (2.1.26) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.26)
- Google providers now support ContentBlock.Multimodal types. (2.1.26) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.26)
- Internal fields are stripped from text content blocks before they are sent to the Anthropic API. (2.1.23) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.23)
- Fixes empty text blocks from streaming chunks missing an index when using Claude models. (2.1.22) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.22)
- Claude Vertex formatting preserves `tool_use` blocks from assistant content and deduplicates them when they also appear in `tool_calls`. (2.1.21) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.21)
- Google models now set the correct _llmType, so code relying on the model type receives the appropriate value. (2.1.17) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.17)
- Google Common and VertexAI provider operations check and propagate abort signals; stream() throws AbortError when aborted, and invoke() supports cancellation during streaming. (2.1.14) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.14)
- Zod-to-Gemini schema conversion now supports nullable object types. (2.1.14) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.14)
- Gemini tools now handle unsupported Zod schema features. (2.1.0) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%402.1.0)

## @langchain/google-common 1

### Security

- Updated the polynomial regex to mitigate a ReDoS vulnerability in SseStream. (1.0.4) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%401.0.4)

### Fixes and other changes

- Fixed compatibility with moduleResolution set to node. (1.0.4) — [release notes](https://github.com/langchain-ai/langchainjs/releases/tag/%40langchain/google-common%401.0.4)