Ollama runtime (CUDA)
ollamaPublished, not hardware-tested
Operator-approved runtime configuration, not GPU-tested.
No ratings yet0 runs
- Min VRAM
- 8 GiB
- Backend
- CUDA · NVIDIA
- Image
- prebuilt
Pick a workload, then a template. Each template lists what it needs and whether it has been tested on real hardware; the machines that fit and their prices come next.
Performance data not yet measured
No template has a measured speed on AmpleRun machines yet, so nothing here is ranked by tokens per second. Compare what each template needs and its status; speed rankings start once benchmarks run on real hosts.
2 templates
Operator-approved runtime configuration, not GPU-tested.
SSH-only classical workload; readiness is the SSH endpoint probe (no HTTP readiness).