Pricing and Billing
Hourly rates, term discounts, and wallet currency — everything before you hit "create".
Pricing model
Everything is priced per hour in NPR. Longer commitments apply the published multipliers to the base hourly rate:
Compute catalog (Nepal market — NPR, VAT-exclusive)
All standard prices below are in NPR and exclude 13% VAT, which is applied at checkout per the compute pricing policy.
For single-node tiers, vCPU, RAM, and disk describe the instance. For multi-node tiers, these resources are configurable. The Nodes column tells you how many separate machines that total is spread over — see Single-Node Instances and Multi-Node Instances for what each layout is good for.
| Tier | Nodes | GPU (VRAM) | vCPU | RAM | Disk | Hourly (NPR) | Daily (NPR) | Weekly (NPR) | Monthly (NPR) |
|---|---|---|---|---|---|---|---|---|---|
| RTX-3090-S | 1 | NVIDIA GeForce RTX 3090 (24 GB) | 4 | 16 GB | 100 GB | 37.80 | 907 | 6,350 | 19,278 |
| RTX-3090-M | 1 | NVIDIA GeForce RTX 3090 (24 GB) | 8 | 24 GB | 200 GB | 52.92 | 1,270 | 8,891 | 26,989 |
| RTX-3090-2X | 2 | 2× NVIDIA GeForce RTX 3090 (48 GB total) | Configurable | Configurable | Configurable | TBD | TBD | TBD | TBD |
| RTX-5080-S | 1 | NVIDIA GeForce RTX 5080 (16 GB) | 8 | 16 GB | 100 GB | 52.92 | 1,270 | 8,891 | 26,989 |
| RTX-5080-M | 1 | NVIDIA GeForce RTX 5080 (16 GB) | 8 | 24 GB | 200 GB | 68.04 | 1,633 | 11,431 | 34,700 |
| RTX-5080-L | 1 | NVIDIA GeForce RTX 5080 (16 GB) | 12 | 32 GB | 300 GB | 83.15 | 1,996 | 13,969 | 42,407 |
| RTX-5090-S | 1 | NVIDIA GeForce RTX 5090 (32 GB) | 8 | 32 GB | 150 GB | 98.27 | 2,358 | 16,509 | 50,118 |
| RTX-5090-M | 1 | NVIDIA GeForce RTX 5090 (32 GB) | 12 | 48 GB | 250 GB | 113.39 | 2,721 | 19,050 | 57,829 |
| RTX-5090-XL | 1 | NVIDIA GeForce RTX 5090 (32 GB) | 16 | 56 GB | 300 GB | 136.07 | 3,266 | 22,860 | 69,396 |
| RTX-5090-XXL | 1 | NVIDIA GeForce RTX 5090 (32 GB) | 16 | 56 GB | 600 GB | 158.75 | 3,810 | 26,670 | 80,963 |
| RTX-5090-2X | 2 | 2× NVIDIA GeForce RTX 5090 (64 GB total) | Configurable | Configurable | Configurable | TBD | TBD | TBD | TBD |
The 2X tiers span two hosts. Their resources are configurable and pricing is to be determined.
Availability status
- RTX 3090 (S, M, 2X) — Available to request
- RTX 5080 (S, M, L) — Available to request
- RTX 5090 (S, M, XL, XXL, 2X) — Available to request
- RTX PRO 6000 — configurations and pricing coming soon
- DGX Spark — configurations and pricing coming soon
Confirm current capacity with ANK before provisioning.
Multi-node (2X) pricing
The RTX-3090-2X and RTX-5090-2X tiers consist of two independent instances on separate hosts, each with one GPU. CPU cores, memory, and storage are configurable, and the hourly rate is to be determined. Contact ANK for a quote and availability.
VRAM is distributed across the two GPUs, rather than combined into one memory space. See Multi-Node Instances for connection and workload details. No early-bird price has been confirmed for these tiers.
Early-bird promotional offers (reference only)
The following promotional prices are published reference figures from ANK product materials. They are VAT-inclusive (unlike the standard catalog above which is VAT-exclusive) and are not automatically active. Contact ANK to confirm current offer status.
Early-bird weekly (20% discount + 15 free compute-hours, up to 27% savings)
| Tier | Standard weekly (incl. VAT) | Early-bird weekly (incl. VAT) |
|---|---|---|
| RTX-3090-S | 7,176 | 5,228 |
| RTX-3090-M | 10,046 | 7,319 |
| RTX-5080-S | 10,046 | 7,319 |
| RTX-5080-M | 12,917 | 9,411 |
| RTX-5080-L | 15,785 | 11,501 |
| RTX-5090-S | 18,656 | 13,592 |
| RTX-5090-M | 21,526 | 15,683 |
| RTX-5090-XL | 25,832 | 18,820 |
| RTX-5090-XXL | 30,137 | 21,957 |
| SPARK-4XL | 22,961 | 16,729 |
| RTX-PRO6000-S | 17,640 | 12,860 |
| RTX-PRO6000-M | 21,168 | 15,430 |
| RTX-PRO6000-L | 24,696 | 18,000 |
| RTX-PRO6000-XL | 28,224 | 20,570 |
Early-bird monthly (30% discount + 7-day free trial, up to 48% savings)
| Tier | Standard monthly (incl. VAT) | Early-bird monthly (incl. VAT) |
|---|---|---|
| RTX-3090-S | 28,704 | 15,069 |
| RTX-3090-M | 40,185 | 21,097 |
| RTX-5080-S | 40,185 | 21,097 |
| RTX-5080-M | 51,667 | 27,125 |
| RTX-5080-L | 63,141 | 33,149 |
| RTX-5090-S | 74,622 | 39,177 |
| RTX-5090-M | 86,104 | 45,205 |
| RTX-5090-XL | 103,326 | 54,246 |
| RTX-5090-XXL | 120,548 | 63,288 |
| SPARK-4XL | 91,845 | 48,218 |
| RTX-PRO6000-S | 70,560 | 37,044 |
| RTX-PRO6000-M | 84,672 | 44,454 |
| RTX-PRO6000-L | 98,784 | 51,862 |
| RTX-PRO6000-XL | 112,896 | 59,270 |
The multi-node tiers (RTX-3090-2X, RTX-5090-2X) are excluded — no promotional figure has been confirmed for them. See Multi-node (2X) pricing.
Storage plans (separate from compute)
Storage is sold separately from compute. The compute tiers above already include a local disk capacity; these storage plans are independent subscriptions and are NOT automatically attached to a compute instance.
Nepal market (NPR, VAT-exclusive)
| Plan | Storage | Monthly (NPR) |
|---|---|---|
| Starter | 20 GB | 1,500 |
| Basic | 50 GB | 2,500 |
| Pro | 100 GB | 5,000 |
| Business | 250 GB | 7,500 |
| Scale | 500 GB | 10,000 |
| Enterprise | 500 GB+ | Custom |
US / International market (USD)
| Plan | Storage | Monthly (USD) | Effective USD/GB | Discount vs. $0.10/GB |
|---|---|---|---|---|
| Starter | 20 GB | $2.00 | $0.100 | — |
| Basic | 50 GB | $4.50 | $0.090 | 10% |
| Pro | 100 GB | $8.00 | $0.080 | 20% |
| Business | 250 GB | $17.50 | $0.070 | 30% |
| Scale | 500 GB | $32.50 | $0.065 | 35% |
| Enterprise | 500 GB+ | Custom | Custom | Custom |
Note: Nepal NPR prices are independently defined market-specific price points. They are NOT USD-to-NPR conversions and must not be replaced by converted values.
Wallet currency
Usage bills in USD. You can top up the wallet in USD or NPR. Glance at the displayed rate and your balance before creating a paid instance.
A cheaper launch sequence
- Top up enough for a short test run.
- Launch a short representative job.
- Confirm access, image behavior, and GPU visibility.
- Only then pick a longer billing plan.
Spending an hour testing before committing to a monthly plan is how you avoid paying for something that doesn't fit. For plan- or invoice-specific questions, Support has you.