RELEASE · MODELS · #2113
AWS open-sources Strands Decider 2B and launches Strands Box sandbox for agent execution
AWS released Strands Decider 2B, a 2-billion-parameter decision-focused model (based on Qwen3.5-2B with a pointer head and rank-16 LoRA) and published the model weights, training data, and scripts. It also announced Strands Box (developer preview), a kernel-enforced sandbox using the Dogwood policy language to control agent execution (shell, Python, tools, filesystem, network); AWS reports ~115 ms median decision latency on an RTX 3090 and 72.3% accuracy on JevBench for a v19 checkpoint.
KEY POINTS
- AWS released Strands Decider 2B, a 2-billion-parameter decision-focused model (based on Qwen3.5-2B with a pointer head and rank-16 LoRA) and published the model weights, training data, and scripts.
- It also announced Strands Box (developer preview), a kernel-enforced sandbox using the Dogwood policy language to control agent execution (shell, Python, tools, filesystem, network); AWS reports ~115 ms median decision latency on an RTX 3090 and 72.3% accuracy on JevBench for a v19 checkpoint.
- These projects give developers a small, open decision model plus a kernel-level sandbox to enforce execution policies, enabling lower-latency local decision-making and stronger runtime controls for AI agents.
WHY IT MATTERS
These projects give developers a small, open decision model plus a kernel-level sandbox to enforce execution policies, enabling lower-latency local decision-making and stronger runtime controls for AI agents.