CloudIndicative GPU & inference pricing
GPU cloud pricing in Nepal
Indicative NVIDIA B300 & inference rates
Indicative rates for planned single-tenant NVIDIA B300 servers in a third-party data centre in Nepal, under NDA, on Himalayan hydropower. No capacity is in service yet; reserved terms from 1 to 36 months will be quoted per engagement.
- Single-tenant servers (planned)
- Indicative pricing
- Platform on a waitlist
Dedicated Bare-Metal GPUs
Planned single-tenant B300 servers with exclusive GPU access — no hypervisor overhead and no noisy neighbors, for training and fine-tuning.
Status: WaitlistThe estimate below is the indicative on-demand rate. Reserved terms from one to thirty-six months will be quoted per engagement — talk to us about a commitment.
Compute Summary
Indicative estimateAI Model Inference
Indicative rates for LLMs and vision models CloudFARM plans to serve. No model is served yet: inference is on a waitlist until serving capacity is connected.
Status: WaitlistLlama 3.3 70B
State-of-the-art open model for reasoning and code generation.
Qwen 2.5 72B
High-performance model with advanced reasoning capabilities.
DeepSeek-R1
Reasoning model with step-by-step thinking capabilities.
FLUX.1 [dev]
Next-gen text-to-image with photorealism and typography.
Indicative Pricing
Indicative on-demand rates for planned capacity. Nothing can be ordered yet; reserved terms will be quoted per engagement.
| GPU Model | On-Demand (indicative) | Monthly (indicative) | Reserved (1-36 months) |
|---|---|---|---|
NVIDIA B300 288GB HBM3e | $18/hr | $12,960/mo | Request a quote |
Frequently Asked Questions
Answers about the GPU cloud platform we are building.
Planning GPU capacity?
Nothing can be ordered yet. Tell us what you need, and we will be in touch as Phase 1 capacity comes online.