Google · 2026-05-01 · major
Google ADK Python v1.32.0 — Native OpenTelemetry Metrics, Anthropic Thinking Blocks, Visual Graph Canvas
Open-source multi-agent framework hits v1.32.0 with native OpenTelemetry agentic metrics, Anthropic thinking block support, visual graph canvas for agent architecture, and a new live UI with session display names.
Google's ADK now ships native OpenTelemetry metrics and visual agent graphs in its 2026 multi-agent framework.
Key specs
| GitHub stars | 19,480 |
|---|
What is it?
The Agent Development Kit (ADK) is Google's open-source Python toolkit for building, evaluating, and deploying multi-agent systems. It launched in April 2025 and has iterated rapidly, reaching v1.32.0 on May 1, 2026 with 19k GitHub stars.
How does it work?
ADK provides a modular runtime where agents are defined as composable nodes connected via typed events. The v1.32.0 release adds native OpenTelemetry integration for agentic metrics — including agent execution time, tool call latency, and LLM cache hit rates — and ships a visual graph canvas in the development UI so engineers can inspect their agent topology. Anthropic thinking-block support lets Claude Opus/Sonnet's extended thinking mode work natively within ADK workflows.
Why does it matter?
Observability has been a persistent gap in multi-agent frameworks. Shipping OpenTelemetry natively means ADK agents can be monitored with the same tooling used for the rest of the stack. The visual graph canvas lowers the debugging barrier for complex agent topologies.
Who is it for?
Python developers building production multi-agent systems with Google Cloud, Vertex AI, or any OpenAI-compatible API.
Try it
pip install google-adk==1.32.0