ServaRICA GPU VPS Deal: Tesla P40 & V100 from $139/mo

Looking for a private, dedicated GPU server to run large language models, fine-tune open-source models or power your own AI tools? ServaRICA has launched a lineup of GPU VPS plans in Montreal, Canada, each with a full NVIDIA Tesla P40 or Tesla V100 card passed through to your server. The GPU is never shared, the cores are dedicated, and the NVIDIA driver comes preinstalled, so you can start running models within minutes. Plans start at just $139/month.

ServaRICA GPU VPS Deal: Annual Billing Savings

Every GPU VPS plan can be billed monthly or annually, and the annual option costs less. The annual price works out to roughly 11 months for the price of 12, so you get one month free on any plan.

Save with annual billing on all six GPU VPS plans (for example, P40 GPU Slice 1 drops from $139/month to about $127.42/month).
7-day refund on first orders, according to ServaRICA.
99.99% network SLA with automatic cash refunds if uptime targets are missed.
Two bandwidth options on every plan: pick what fits your workload at checkout.

Why Choose a GPU VPS from ServaRICA?

Full GPU passthrough: the whole card is dedicated to your VPS, with no sharing and no noisy neighbors.
Dedicated cores and NVMe storage: AMD EPYC Gen 2 or Intel Xeon Gold Gen 2 cores, running on KVM with Proxmox.
Linux or Windows: choose Ubuntu or Windows Server 2022 at deployment, with GPU drivers already installed.
Privacy first: keep your prompts and documents on your own server, and switch models whenever you like.
Proven AI performance: ServaRICA’s own tests (October 2026) show a single Tesla V100 reaching about 72 to 79 tokens per second when generating text with the 125B Qwen3.8-Flash-Next model on Strata, and a single Tesla P40 reaching about 39 to 41 tokens per second. Smaller models run faster: up to 154 tokens per second on a V100 and up to 73 tokens per second on a P40 with llama.cpp.

GPU VPS Plans and Pricing

All prices are in USD and billed monthly unless noted. Annual prices are shown for comparison. Every plan includes full GPU passthrough, dedicated cores, NVMe storage, a preinstalled driver and a choice between two bandwidth options.

P40 GPU Slice 1 – 1x Tesla P40 (24 GB VRAM)

GPU: 1x NVIDIA Tesla P40, 24 GB VRAM
CPU: 10 dedicated cores
RAM: 40 GB
Storage: 500 GB NVMe
Price: $139/month or $1,529/year (about $127.42/month)
Order Now

V100 GPU Slice 1 – 1x Tesla V100 (32 GB HBM2)

GPU: 1x NVIDIA Tesla V100, 32 GB HBM2
CPU: 8 dedicated cores
RAM: 35 GB
Storage: 400 GB NVMe
Price: $179/month or $1,969/year (about $164.08/month)
Order Now

P40 GPU Slice 2 – 2x Tesla P40 (48 GB VRAM)

GPU: 2x NVIDIA Tesla P40, 48 GB VRAM in total
CPU: 20 dedicated cores
RAM: 80 GB
Storage: 1 TB NVMe
Price: $278/month or $3,058/year (about $254.83/month)
Order Now

V100 GPU Slice 2 – 2x Tesla V100 (64 GB HBM2)

GPU: 2x NVIDIA Tesla V100, 64 GB HBM2 in total
CPU: 16 dedicated cores
RAM: 70 GB
Storage: 800 GB NVMe
Price: $358/month or $3,938/year (about $328.17/month)
Order Now

P40 GPU Slice 3 – 3x Tesla P40 (72 GB VRAM)

GPU: 3x NVIDIA Tesla P40, 72 GB VRAM in total
CPU: 30 dedicated cores
RAM: 120 GB
Storage: 1.5 TB NVMe
Price: $417/month or $4,587/year (about $382.25/month)
Order Now

V100 GPU Slice 3 – 3x Tesla V100 (96 GB HBM2)

GPU: 3x NVIDIA Tesla V100, 96 GB HBM2 in total
CPU: 24 dedicated cores
RAM: 105 GB
Storage: 1.2 TB NVMe
Price: $537/month or $5,907/year (about $492.25/month)
Order Now

Bandwidth Options (All GPU VPS Plans)

Option 1: Unlimited transfer on a 250 Mbps port that grows by 1 Mbps per day, up to 1 Gbps
Option 2: 10 Gbps port with 100 TB of monthly transfer (speed drops to about 10 Mbps after the cap)

About ServaRICA

ServaRICA is a Canadian hosting provider that has been in business since 2010. It began when founder Hani could not find an affordable, fully Canadian-owned host for his own project, so the team bought hardware and started hosting for others. Today the company operates its own servers, routers and ASN in Montreal, and describes itself as Canadian-owned and not a reseller. Its services include KVM VPS, NVMe storage plans, GPU VPS, bare-metal dedicated servers and cPanel shared hosting. The company is operated by Rica Web Services Inc.

Founded: 2010
Headquarters: 350 Rue de Louvain O, Suite 203C, Montreal, QC H2N 2E8, Canada
Support: [email protected]

Payment Information

PayPal
Alipay
Credit cards (Visa, Mastercard and other major cards)
Cryptocurrencies (Bitcoin, Ethereum and more)
Bank transfer and mailed checks (for amounts over $200)
Currencies: USD and CAD

Network Information

Server Location: Montreal, Quebec, Canada, close to the US East Coast
Test IPv4: 162.250.188.171
Test IPv6: 2602:ffd5:1:141::1
Network: Owned and operated by ServaRICA, multi-homed with multiple upstream carriers and redundant routers, switches and power
Power: 100% renewable energy from Hydro-Québec
SLA: 99.99% network uptime guarantee with automatic cash refunds if it is missed
Typical latency: single-digit milliseconds to New York, Boston and Washington DC, under 10 ms to Toronto, 60 to 80 ms to the US West Coast and 70 to 100 ms to major European cities