Targon

Targon, a platform from Manifold Labs, provides on-demand GPU and CPU computing for AI training, inference, code execution, and agent development. Customers can deploy virtual machines with SSH access and up to eight GPUs, or choose bare-metal servers for direct access to the hardware. Targon also offers isolated sandboxes for running code and agents, plus reserved GPU capacity arranged through its team. Services are available with hourly pricing, and users can deploy through the console or command-line interface.

For workloads requiring additional protection, eligible virtual machines run in Intel TDX confidential computing environments. These use hardware-isolated trust domains, encrypted memory, and boot-time attestation before a disk key is released. On supported hardware, NVIDIA Confidential Computing also encrypts transfers between a virtual machine and its GPU. Targon provides developer tools including a CLI, Python SDK, API keys, organization-owned service tokens, and SSH key management. TypeScript support is listed as coming soon.