Daytona GPU Sandboxes
Linux container sandboxes with one or more NVIDIA or AMD GPUs attached, on on-demand or reclaimable spot capacity, for AI agents that write, run and profile GPU code such as CUDA kernels; priced per GPU-hour.
Find alternatives to Daytona GPU Sandboxes
What it does
- GPU cloud
- Agent orchestration
Sources
- Pricing
-
Page text fetched 2026-11-16 with curl; page quote: "Nvidia H100 $2.27/h". The pricing table lists GPU rates per hour for preemptible and on-demand capacity (Nvidia B300 through RTX 4090, and AMD MI355X) beside the compute rates; "All billing is calculated per second."
- Description
-
Official documentation · 16 Nov 2026
Page text fetched 2026-11-16 with curl; page quote: "Daytona provides GPU sandboxes for AI agents doing GPU-accelerated work, such as writing and profiling CUDA kernels, 3D and graphics workloads, and other GPU-accelerated compute.".
- What it does
-
Official documentation · 16 Nov 2026
Page text fetched 2026-11-16 with curl; page quote: "Daytona provides GPU sandboxes for AI agents doing GPU-accelerated work, such as writing and profiling CUDA kernels, 3D and graphics workloads, and other GPU-accelerated compute.".
- Sold within
-
Page text fetched 2026-11-16 with curl; page quote: "Daytona lets you spin up sandboxes in milliseconds and shut them down just as fast. Use what you need, when you need it.". GPU rates sit in the same pay-as-you-go table as vCPU, memory and storage for Daytona Sandboxes, the same shape as the recorded daytona-computer-use-sandboxes.
- Name
-
Official documentation · 16 Nov 2026
Page text fetched 2026-11-16 with curl; page quote: "Daytona provides GPU sandboxes for AI agents doing GPU-accelerated work, such as writing and profiling CUDA kernels, 3D and graphics workloads, and other GPU-accelerated compute.".
- URL
-
Official documentation · 16 Nov 2026
Page text fetched 2026-11-16 with curl; page quote: "Daytona provides GPU sandboxes for AI agents doing GPU-accelerated work, such as writing and profiling CUDA kernels, 3D and graphics workloads, and other GPU-accelerated compute.".
- Deployment
-
Official documentation · 16 Nov 2026
Page text fetched 2026-11-16 with curl; GPU sandboxes are created through the Daytona service ("Create a GPU sandbox") on "the shared region that runs GPU sandboxes on shared capacity"; no self-hosted option is described for GPU sandboxes.
Also from Daytona
-
Daytona Sandboxes Infrastructure service
Elastic sandboxes for running AI-generated code, with process execution, file system, Git and LSP APIs, snapshots, volumes and a declarative image builder (Dockerfile and Docker Compose supported); pay-as-you-go with $200 in free compute, vCPU $0.0504/h, memory $0.0162/GiB/h.
-
Daytona Computer Use Sandboxes Infrastructure service
Sandboxes with a full desktop for computer-use agents on Linux, macOS and Windows (Windows at $0.0858 per vCPU-hour).
Something wrong here? Send a correction — quote this product id: daytona-gpu-sandboxes.