DeepSeek Details DSec Sandbox Platform for Agentic RL Training
DeepSeek Details DSec Elastic Compute: Agentic-Training Sandboxes at ~3M/Day, 380K+ Concurrent
The throughput figures reveal the infrastructure scale behind agentic RL training, which most labs keep undisclosed. Teams building agent training environments or RL pipelines can use the reported concurrency and creation-rate numbers as a reference point when sizing their own sandbox fleets. The arXiv report with Wenfeng Liang as co-author signals this is a company-level engineering priority, not a side project.
DeepSeek published details of DSec, its Elastic Compute sandbox platform used for agentic reinforcement learning training.
The platform runs roughly 3 million sandboxes per day, with peak concurrency above 380,000 and sandbox creation rates exceeding 5,000 per second.
A technical report was released on arXiv as 2609.22978, with DeepSeek founder Wenfeng Liang listed among the co-authors.
DeepSeek Details DSec Elastic Compute: Agentic-Training Sandboxes at ~3M/Day, 380K+ Concurrent
DeepSeek detailed DSec, its Elastic Compute sandbox platform for agentic RL, reporting about 3M sandboxes/day, peak concurrency above 380k and create rates over 5,000/s, with Wenfeng Liang among co-authors on arXiv 2609.22978.
This report is categorized under AITOP's "Papers" desk, focusing on China's AI industry and domestic foundation model ecosystem. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison and AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include DeepSeek, agentic RL, DSec, reinforcement learning.
Primary source verification: published directly by pandaily with first-party authority.
