Moonshot’s Kimi K3 report details a microVM sandbox system for agentic RL
stochasticchasm · x · 2026-07-28
The thread highlights a section from Kimi K3’s report on sandbox infrastructure for post-training and evaluation.
Moonshot says it uses multiple sandbox runtimes, including containers, GPU sandboxes, and a new microVM-based system called AgentENV. The design goals are: high-fidelity isolation, flexible life cycles for agentic RL, and high-density scaling.
Notable features include:
- Pause/resume for sandboxes that may spend up to 98% of their lifetime waiting on model inference
- Fork to branch from an exact snapshot without stopping the original sandbox
- Snapshot support for recovery
- Fast checkpoint/resume latencies of 133 ms / 49 ms
- Scale data from the report: 51,219,741 sandboxes across 1,505,678 images
More from Infra
- Optimizing K8s Resource Requests Yields 9x Speedup for Whisper Workloads — anacondainc · 2026-07-28
- Linear-attention hybrids may need finer caching for long prompts and workflows — stochasticchasm · 2026-07-28
- A research-agent ranking of 8 stock-data MCP servers puts Equibles first — DanielAPO · 2026-07-28
- Underlayer Electrons Aggravate Stochastic Defectivity in EUV Lithography — CatAstro_Piyush · 2026-07-28
- SkyPilot says serving Kimi K3 needs multi-node inference and a full stack — skypilot_org · 2026-07-28
- A 6600XT user says ROCm feels slower than Vulkan in LM Studio — InsideYork · 2026-07-28