Strands Agents 1.65.0 released
A feature release adding async agent builders, improved model handling, and expanded integrations including SageMaker decision models and configurable shell timeouts.
What changed in 1.65.0
- Async agent builders now report content-filtered and refused children as tool errors.
- Added SageMakerDecisionModel for System One models on SageMaker AI endpoints.
- Shell tool default timeout is now configurable.
- Injected text written into history as an injected block; improved MCP search with alternate phrasings and field-weighted ranking.
- Support for anthropic SDK versions 1.x and 2.x; added Claude Fable 5.2, Mythos 5.2, and GPT-6 Cyber model info.
Check before you update
- Offload conditions now decide granularity (context feature change).
- Todos dropped from default built-in plugins in harness.
- Injected text handling changed to use InjectedBlock format.
Summarised by AgentGid from AWS's official release notes on GitHub, as reported there; read the original before relying on it. Price, status and release dates come from our daily data.
Strands Agents at a glance
| What it is | AWS's open-source SDK for building and running AI agents in Python and TypeScript: a model-driven agent loop that runs in your own process with no hosted control plane; the same repository holds a harness, a CLI, MCP su… |
|---|---|
| Latest version | 1.65.0 is the latest stable release we track. |
| Release pace | 22 stable releases in the last 30 days; 1.65.0 came the same day as the previous one |
| Price | Free (OSS). Strands Agents is open source and free to use; you pay only for the models and infrastructure it runs on. (checked Nov 4, 2026) |
| Downloads, last 7 days | 8.9M up 0.8% |
| Gid Score | 55 #10 in Agent frameworks and SDKs |