Mini Activations Narrow Frontier Model Gaps
A sparse 229.9B-parameter MoE activates 9.8B parameters per token and pairs agent-native data with RL for coding, cowork, and reasoning tasks.
Jul 31, 20265 min2605.26494
2 articles on SOTA Papers
A sparse 229.9B-parameter MoE activates 9.8B parameters per token and pairs agent-native data with RL for coding, cowork, and reasoning tasks.
Workflow-GYM evaluates 338 tasks across 56 virtual-machine software environments, where the best model averages 30.67% success.
