Google Gen AI SDK v2.33.0
Google Gen AI SDK v2.33.0 adds Python 3.15 support, new config fields ( store , labels ), multimodal function result contents, and various client bug fixes.
Why this signal matters
Google Gen AI SDK v2.33.0 introduces several new features and bug fixes. Key features include adding store to GenerateContentConfig (restricted to the Gemini Developer API) and labels to LiveConnectConfig . Multimodal responses in tool calls are expanded by including AudioContent , DocumentContent , and VideoContent in FunctionResultSubContent . The release adds support for Python 3.15 and client-side configuration for httpx2 via GOOGLE GENAI HTTP CLIENT . Bug fixes address skipping mTLS URL rewriting for multi-regional .rep.googleapis.com endpoints, honoring GOOGLE API USE MTLS ENDPOINT in Google Auth sessions, respecting explicit zero delay fields in HttpRetryOptions , and properly honoring HttpRetryOptions on aiohttp connection errors during async requests. Documentation was also updated regarding unenforced JSON Schema keywords on Interactions schema fields.
Actionable summary
Google Gen AI SDK v2.33.0 adds store to GenerateContentConfig (Gemini Developer API only), labels to LiveConnectConfig , multimodal sub-content types to FunctionResultSubContent , and httpx2 support via GOOGLE GENAI HTTP CLIENT . It also addresses mTLS URL rewriting, explicit zero delays in HttpRetryOptions , and aiohttp connection error retries.
- Agent usefulness
- 77/100
- Confidence
- 96%
- Canonical data
- JSON + Markdown
What builders should check
- Upgrade google-genai SDK to v2.33.0 if using new configuration fields or Python 3.15.
- Verify async retry behavior if relying on HttpRetryOptions with aiohttp connection errors or zero-delay configurations.
- Adopt AudioContent, DocumentContent, or VideoContent in FunctionResultSubContent if returning multimodal outputs from tool calls.