{"type":"post","id":"nvidia-releases-open-source-snapshot-for-checkpoint-and-57b093","slug":"nvidia-releases-open-source-snapshot-for-checkpoint-and-57b093","title":"NVIDIA releases open-source Snapshot for checkpoint and restore of GPU inference workloads on Kubernetes","description":"NVIDIA has released Snapshot, an open-source Kubernetes component that checkpoints initialized GPU workloads and restores them without repeating model loading, compilation, or CUDA graph capture.","retrieval_nugget":"NVIDIA has released Snapshot, an open-source Kubernetes component that checkpoints initialized GPU workloads and restores them without repeating model loading, compilation, or CUDA graph capture. NVIDIA reports restore times of 3.6 to 42.7 seconds across tested configurations, compared with cold starts of up to 19 minutes. The practical promise is faster scaling and recovery for compatible inference services.","published_at":"2026-11-10","updated_at":"2026-11-10","record_date":"2026-11-10","date_kind":"published_at","topics":["inference"],"entities":["NVIDIA","NVIDIA Snapshot","NVIDIA Dynamo","vLLM","SGLang","TensorRT LLM","Kubernetes"],"related_patterns":["harness-architecture-outlives-model-choice"],"novelty_label":"new_fact","what_changed":"NVIDIA released Snapshot, an open-source Kubernetes system that checkpoints and restores initialized GPU inference workloads to bring replicas online in seconds.","prior_refs":["https://newruntime.com/posts/nvidia-s-gpu-sharing-software-is-now-free-8c9a45/"],"source_url":"https://developer.nvidia.com/blog/start-and-restore-gpu-workloads-in-seconds-with-nvidia-snapshot","source_urls":["https://developer.nvidia.com/blog/start-and-restore-gpu-workloads-in-seconds-with-nvidia-snapshot"],"schema_version":"newruntime-agent-readable-v0.2","stable_id":"post:nvidia-releases-open-source-snapshot-for-checkpoint-and-57b093","status":"published","visuals":[],"editor":{"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]},"routes":{"html":"https://newruntime.com/posts/nvidia-releases-open-source-snapshot-for-checkpoint-and-57b093/","markdown":"https://newruntime.com/posts/nvidia-releases-open-source-snapshot-for-checkpoint-and-57b093.md","json":"https://newruntime.com/posts/nvidia-releases-open-source-snapshot-for-checkpoint-and-57b093.json"},"markdown_url":"https://newruntime.com/posts/nvidia-releases-open-source-snapshot-for-checkpoint-and-57b093.md","next_reads":[{"type":"topic","path":"/topics/inference/","url":"https://newruntime.com/topics/inference/","title":"Inference","reason":"Explore the inference topic hub.","media_type":"text/html","markdown_url":"https://newruntime.com/topics/inference.md","json_url":"https://newruntime.com/topics/inference.json"},{"type":"collection","path":"/posts/","url":"https://newruntime.com/posts/","title":"Field Notes","reason":"Open the latest editorial analysis.","media_type":"text/html","json_url":"https://newruntime.com/posts.json"},{"type":"collection","path":"/sources/","url":"https://newruntime.com/sources/","title":"Sources","reason":"See the original sources this archive cites.","media_type":"text/html"}]}