# What changed in google-cloud-aiplatform

> 22 changes in google-cloud-aiplatform 1 to 2 from its official release notes, upgrade guides and advisories: 2 security, 7 new APIs and options. Each one cited. Sources checked 2026-10-18. HTML: https://unstale.sh/pypi/google-cloud-aiplatform/

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

## google-cloud-aiplatform 2

### New APIs and options

- Add agent_card to AgentRuntimeConfig for runtimes.create and update. (2.5.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.5.0)
- Agent Engine create and update requests now surface the LRO name. (2.4.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.4.0)
- Evals support cross-region routing. (2.4.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.4.0)
- Adds a method to load an EvaluationDataset from an ADK eval set. (2.4.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.4.0)
- Evals support per-step judges. (2.4.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.4.0)
- LiveClientSetup now includes labels. (2.3.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.3.0)
- The agentplatform genai SDK adds serving profiles support. (2.3.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.3.0)

### Fixes and other changes

- Display N/A for missing summary scores in the comparison report. (2.5.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.5.0)
- Treat NaN and array cells as missing in eval DataFrames. (2.5.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.5.0)
- Send list and dict context to the evaluation service. (2.5.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.5.0)
- Raise google.genai.errors.APIError when Agent Engine create or update LROs fail. (2.5.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.5.0)
- Back off only on retryable errors during inference and agent-run retries. (2.5.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.5.0)
- agent_engines now allows opentelemetry-instrumentation-google-genai 1.x. (2.5.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.5.0)
- Pass client credentials and http_options to per-thread runtime clients. (2.5.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.5.0)
- Fix run_inference for local ADK agents and remote runtime N+1. (2.5.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.5.0)
- Save Gemini agent run_inference results to dest. (2.5.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.5.0)
- Show intermediate_events as a conversation trace in show(). (2.5.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.5.0)
- Session events are replayed into a scratch session, which is deleted afterwards. (2.4.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.4.0)
- async_stream_query is non-blocking by using the async execution client. (2.3.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.3.0)
- clone() keeps the AdkApp subclass, plugins, and credential_service_builder. (2.3.0) — [release notes](https://github.com/googleapis/python-aiplatform/releases/tag/v2.3.0)

## google-cloud-aiplatform 1

### Security

- Google Cloud Vertex AI has a a vulnerability involving predictable bucket naming (1.133.0) — [security advisory](https://github.com/advisories/GHSA-WH2J-26J7-9728)
- Google Cloud Vertex AI SDK affected by Stored Cross-Site Scripting (XSS) (1.131.0) — [security advisory](https://github.com/advisories/GHSA-QV8J-HGPC-VRQ8)
