Google shipped ADK Python v2.10.0 on 2026-09-25 (google-adk 2.10.0): experimental skill lifecycle (ADK_ENABLE_SKILL_LIFECYCLE=1) with ephemeral lifecycles and active skill limits; MongoDB toolset for vector/hybrid search; evaluation metrics for duration, tokens, and model-call counts; better OpenAI reasoning model parameter adaptation and reasoning-token reporting. BigQuery protected write mode and instruction templating tighten with breaking changes.
Key Takeaways
- ✓Shipped v2.10.0 — pip install google-adk==2.10.0; docs at adk.dev
- ✓Experimental skill lifecycle via ADK_ENABLE_SKILL_LIFECYCLE=1
- ✓MongoDB toolset for in-flow vector/hybrid search
- ✓Eval metrics: duration, tokens, model-call counts
- ✓Breaking: tighter BigQuery protected writes; OpenAIResponsesLlm uses effort not thinking_config; empty eval sets raise ValueError
Finished reading? Explore benchmark rankings & pricing
Real-world SWE-bench scores & $20/mo vs API cost break-even calculator
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Core Background & Industry Pain Points
Teams need one Agent runtime to govern Skill resources, call vector DBs, and quantify eval duration/tokens. Always-on skills bloat tools/context; DB search needs glue; evals lacked first-class cost/latency counters; OpenAI reasoning params and BigQuery protected writes needed stricter defaults.
Architecture Highlights & Internals
v2.10.0 adds experimental skill lifecycle (ADK_ENABLE_SKILL_LIFECYCLE=1), a MongoDB toolset for vector/hybrid search, and eval metrics for duration/tokens/model calls. OpenAI reasoning adapts via effort (not thinking_config); BigQuery protected writes tighten around anonymous-dataset dry-runs.
Authoritative Benchmarks & Measured Scores
No public leaderboard ships—use the new duration/token/model-call metrics on your eval sets and compare skill-lifecycle resource use. Do not invent arena scores.
Developer Hands-on Guide
pip install -U google-adk==2.10.0 (PyPI); docs at adk.dev / Skills / Evaluate. See the release.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.