Vultr Cloud Pricing Complete Guide: How Much Does Vultr Actually Cost? Cloud Compute, High Frequency, GPU, and Bare Metal Plans Compared — With Free Credit Codes, Hidden Fees Breakdown, and Plan Selection Tips
If you've ever stared at a cloud provider's pricing page and felt your eyes glaze over, you're in good company. Vultr cloud pricing looks deceptively simple at first glance — a tidy grid of monthly numbers — but the moment you start adding bandwidth, snapshots, backups, GPU hours, and reserved contracts into the math, the picture gets murky fast. This guide walks through everything I found while digging into Vultr's pricing structure, from the $2.50/month entry-level instance to the $13,000/month H100 GPU cluster, including the offers currently live, the fees that catch people off guard, and which plan actually makes sense for which workload.
Before diving into specific plans, it helps to understand the bones of how Vultr charges. Most Vultr Cloud Compute products use a 730-hour monthly billing model — meaning you pay an hourly rate that's capped at 730 hours per month, so even if you run a server 24/7, you'll never pay more than the listed monthly price. That's friendlier than pure per-second billing for steady-state workloads.
Pricing is region-based, which means the same plan can cost slightly different amounts depending on which of the 33 global data centers you deploy into. Bandwidth is bundled into each plan (anywhere from 0.5 TB on the smallest Regular Performance instance up to 25 TB on big GPU and bare metal nodes), and every account also gets a pooled 2 TB/month free egress allowance across all instances. Go over that, and you're looking at $0.01 per GB in overage fees — a number that quietly shows up on a lot of bills.
The pricing page itself is the source of truth, and it's worth bookmarking: 👉 View the complete Vultr pricing grid.
Cloud Compute is what most people picture when they think "Vultr VPS." These run on shared vCPUs and cover the broad middle of workloads: blogs, CMS sites, dev/test boxes, small databases, low-traffic web apps. Vultr splits this category into three sub-tiers based on the underlying hardware.
Powered by previous-generation Intel CPUs and regular SSDs. This is the cheapest entry point, suitable for personal blogs and low-traffic sites. The $2.50/month IPv6-only plan is the lowest sticker price on the entire platform.
| Plan | vCPU | RAM | Storage | Bandwidth | Monthly | Hourly |
|---|---|---|---|---|---|---|
| Regular 1 vCPU (IPv6 only) | 1 | 0.5 GB | 10 GB | 0.5 TB | $2.50 | $0.004 |
| Regular 1 vCPU | 1 | 0.5 GB | 10 GB | 0.5 TB | $3.50 | $0.005 |
| Regular 1 vCPU | 1 | 1 GB | 25 GB | 1 TB | $5 | $0.007 |
| Regular 1 vCPU | 1 | 2 GB | 55 GB | 2 TB | $10 | $0.015 |
| Regular 2 vCPU | 2 | 2 GB | 65 GB | 3 TB | $15 | $0.022 |
| Regular 2 vCPU | 2 | 4 GB | 80 GB | 3 TB | $20 | $0.030 |
| Regular 4 vCPU | 4 | 8 GB | 160 GB | 4 TB | $40 | $0.060 |
| Regular 6 vCPU | 6 | 16 GB | 320 GB | 5 TB | $80 | $0.119 |
| Regular 8 vCPU | 8 | 32 GB | 640 GB | 6 TB | $160 | $0.238 |
| Regular 16 vCPU | 16 | 64 GB | 1280 GB | 10 TB | $320 | $0.476 |
| Regular 24 vCPU | 24 | 96 GB | 1600 GB | 15 TB | $640 | $0.952 |
👉 Deploy a Regular Performance instance
A step up — newer AMD EPYC or Intel Xeon CPUs paired with NVMe SSD. Best for dev/test environments and small databases where you want consistent disk I/O. AMD and Intel variants are priced identically.
| Plan | vCPU | RAM | Storage (NVMe) | Bandwidth | Monthly | Hourly |
|---|---|---|---|---|---|---|
| High Perf 1 vCPU | 1 | 1 GB | 25 GB | 2 TB | $6 | $0.009 |
| High Perf 1 vCPU | 1 | 2 GB | 50 GB | 3 TB | $12 | $0.018 |
| High Perf 2 vCPU | 2 | 2 GB | 60 GB | 4 TB | $18 | $0.027 |
| High Perf 2 vCPU | 2 | 4 GB | 100 GB | 5 TB | $24 | $0.036 |
| High Perf 4 vCPU | 4 | 8 GB | 180 GB | 6 TB | $48 | $0.071 |
| High Perf 4 vCPU | 4 | 12 GB | 260 GB | 7 TB | $72 | $0.107 |
| High Perf 8 vCPU | 8 | 16 GB | 350 GB | 8 TB | $96 | $0.143 |
| High Perf 12 vCPU | 12 | 24 GB | 500 GB | 12 TB | $144 | $0.214 |
👉 Get started with High Performance Cloud Compute
Built on 3 GHz+ Intel Xeon CPUs with NVMe SSD. These are the go-to for content management systems and small game servers where single-thread speed matters more than core count.
| Plan | vCPU | RAM | Storage (NVMe) | Bandwidth | Monthly | Hourly |
|---|---|---|---|---|---|---|
| HF 1 vCPU | 1 | 1 GB | 32 GB | 1 TB | $6 | $0.009 |
| HF 1 vCPU | 1 | 2 GB | 64 GB | 2 TB | $12 | $0.018 |
| HF 2 vCPU | 2 | 2 GB | 80 GB | 3 TB | $18 | $0.027 |
| HF 2 vCPU | 2 | 4 GB | 128 GB | 3 TB | $24 | $0.036 |
| HF 3 vCPU | 3 | 8 GB | 256 GB | 4 TB | $48 | $0.071 |
| HF 4 vCPU | 4 | 16 GB | 384 GB | 5 TB | $96 | $0.143 |
| HF 6 vCPU | 6 | 24 GB | 448 GB | 6 TB | $144 | $0.214 |
| HF 8 vCPU | 8 | 32 GB | 512 GB | 7 TB | $192 | $0.286 |
| HF 12 vCPU | 12 | 48 GB | 768 GB | 8 TB | $256 | $0.381 |
👉 Explore High Frequency Compute plans
These instances run on fully dedicated vCPUs — no noisy neighbors. Vultr positions them for production workloads where consistent CPU performance is non-negotiable. The new VX1™ line in this category is what Vultr has been pushing hardest in marketing, claiming up to 82% better performance per dollar compared to its older Optimized Cloud Compute plans and 33% better cost efficiency per vCPU versus hyperscaler cost-optimized tiers.
The balanced default — equal-ish portions of CPU, RAM, and NVMe. Good starting point for web/app servers, e-commerce, game servers, streaming, API serving, and relational databases.
| Plan | vCPU | RAM | Storage (NVMe) | Bandwidth | Monthly | Hourly |
|---|---|---|---|---|---|---|
| GP 1 vCPU | 1 | 4 GB | 30 GB | 4 TB | $30 | $0.045 |
| GP 2 vCPU | 2 | 8 GB | 50 GB | 5 TB | $60 | $0.089 |
| GP 4 vCPU | 4 | 16 GB | 80 GB | 6 TB | $120 | $0.179 |
| GP 8 vCPU | 8 | 32 GB | 160 GB | 7 TB | $240 | $0.357 |
| GP 16 vCPU | 16 | 64 GB | 320 GB | 8 TB | $480 | $0.714 |
| GP 24 vCPU | 24 | 96 GB | 480 GB | 9 TB | $720 | $1.071 |
| GP 32 vCPU | 32 | 128 GB | 640 GB | 9 TB | $960 | $1.429 |
| GP 40 vCPU | 40 | 160 GB | 768 GB | 10 TB | $1,200 | $1.786 |
| GP 64 vCPU | 64 | 192 GB | 960 GB | 11 TB | $1,920 | $2.857 |
| GP 96 vCPU | 96 | 256 GB | 1280 GB | 12 TB | $3,840 | $5.714 |
Skews heavy on CPU relative to RAM and storage. Aimed at video encoding, batch processing, CI/CD, HPC, ad serving, and analytics.
| Plan | vCPU | RAM | Storage (NVMe) | Bandwidth | Monthly | Hourly |
|---|---|---|---|---|---|---|
| CPU Opt 1 vCPU | 1 | 2 GB | 25 GB | 4 TB | $28 | $0.042 |
| CPU Opt 2 vCPU (50 GB) | 2 | 4 GB | 50 GB | 5 TB | $40 | $0.060 |
| CPU Opt 2 vCPU (75 GB) | 2 | 4 GB | 75 GB | 5 TB | $45 | $0.067 |
| CPU Opt 4 vCPU (75 GB) | 4 | 8 GB | 75 GB | 6 TB | $80 | $0.119 |
| CPU Opt 4 vCPU (150 GB) | 4 | 8 GB | 150 GB | 6 TB | $90 | $0.134 |
| CPU Opt 8 vCPU (150 GB) | 8 | 16 GB | 150 GB | 7 TB | $160 | $0.238 |
| CPU Opt 8 vCPU (300 GB) | 8 | 16 GB | 300 GB | 7 TB | $180 | $0.268 |
| CPU Opt 16 vCPU (300 GB) | 16 | 32 GB | 300 GB | 8 TB | $320 | $0.476 |
| CPU Opt 16 vCPU (500 GB) | 16 | 32 GB | 500 GB | 8 TB | $360 | $0.536 |
| CPU Opt 32 vCPU (500 GB) | 32 | 64 GB | 500 GB | 9 TB | $640 | $0.952 |
| CPU Opt 32 vCPU (1000 GB) | 32 | 64 GB | 1000 GB | 10 TB | $720 | $1.071 |
Lots of RAM, modest CPU and storage. Built for in-memory databases (Memcached), MySQL, real-time analytics, and anything that wants to keep big datasets in RAM.
| Plan | vCPU | RAM | Storage (NVMe) | Bandwidth | Monthly |
|---|---|---|---|---|---|
| Mem Opt 1 vCPU | 1 | 8 GB | 50 GB | 5 TB | $40 |
| Mem Opt 2 vCPU (100 GB) | 2 | 16 GB | 100 GB | 6 TB | $80 |
| Mem Opt 2 vCPU (200 GB) | 2 | 16 GB | 200 GB | 6 TB | $100 |
| Mem Opt 2 vCPU (400 GB) | 2 | 16 GB | 400 GB | 6 TB | $125 |
| Mem Opt 4 vCPU (200 GB) | 4 | 32 GB | 200 GB | 8 TB | $160 |
| Mem Opt 4 vCPU (400 GB) | 4 | 32 GB | 400 GB | 8 TB | $195 |
| Mem Opt 4 vCPU (800 GB) | 4 | 32 GB | 800 GB | 8 TB | $250 |
| Mem Opt 8 vCPU (400 GB) | 8 | 64 GB | 400 GB | 9 TB | $320 |
| Mem Opt 8 vCPU (800 GB) | 8 | 64 GB | 800 GB | 9 TB | $390 |
| Mem Opt 8 vCPU (1600 GB) | 8 | 64 GB | 1600 GB | 9 TB | $500 |
| Mem Opt 16 vCPU (800 GB) | 16 | 128 GB | 800 GB | 10 TB | $640 |
| Mem Opt 16 vCPU (1600 GB) | 16 | 128 GB | 1600 GB | 10 TB | $785 |
| Mem Opt 16 vCPU (3200 GB) | 16 | 128 GB | 3200 GB | 10 TB | $1,000 |
| Mem Opt 24 vCPU (1200 GB) | 24 | 192 GB | 1200 GB | 11 TB | $960 |
| Mem Opt 24 vCPU (2400 GB) | 24 | 192 GB | 2400 GB | 11 TB | $1,175 |
| Mem Opt 24 vCPU (4800 GB) | 24 | 192 GB | 4800 GB | 11 TB | $1,500 |
| Mem Opt 32 vCPU (1600 GB) | 32 | 256 GB | 1600 GB | 12 TB | $1,280 |
| Mem Opt 32 vCPU (3200 GB) | 32 | 256 GB | 3200 GB | 12 TB | $1,565 |
Big NVMe SSD footprints paired with balanced CPU/RAM. Aimed at large NoSQL databases (Cassandra, MongoDB) and high-frequency OLTP workloads.
| Plan | vCPU | RAM | Storage (NVMe) | Bandwidth | Monthly |
|---|---|---|---|---|---|
| Stor Opt 1 vCPU | 1 | 8 GB | 150 GB | 4 TB | $75 |
| Stor Opt 2 vCPU (320 GB) | 2 | 16 GB | 320 GB | 6 TB | $125 |
| Stor Opt 2 vCPU (480 GB) | 2 | 16 GB | 480 GB | 6 TB | $155 |
| Stor Opt 4 vCPU (640 GB) | 4 | 32 GB | 640 GB | 7 TB | $250 |
| Stor Opt 4 vCPU (960 GB) | 4 | 32 GB | 960 GB | 7 TB | $310 |
| Stor Opt 8 vCPU (1280 GB) | 8 | 64 GB | 1280 GB | 8 TB | $500 |
| Stor Opt 8 vCPU (1920 GB) | 8 | 64 GB | 1920 GB | 8 TB | $620 |
| Stor Opt 16 vCPU (2560 GB) | 16 | 128 GB | 2560 GB | 9 TB | $1,000 |
| Stor Opt 16 vCPU (3840 GB) | 16 | 128 GB | 3840 GB | 9 TB | $1,240 |
| Stor Opt 24 vCPU (3840 GB) | 24 | 192 GB | 3840 GB | 10 TB | $1,500 |
| Stor Opt 24 vCPU (5760 GB) | 24 | 192 GB | 5760 GB | 10 TB | $1,850 |
| Stor Opt 32 vCPU | 32 | 256 GB | 5760 GB | 12 TB | $2,000 |
👉 Compare Optimized Cloud Compute plans
Vultr has gone aggressive on GPU pricing, and the lineup now spans from fractional GPUs up to 8-GPU H100 nodes. GPU pricing is structured differently from regular compute — many flagship SKUs (H100, GH200) are quoted on 36-month, 100% prepaid reserved instance contracts, with the displayed hourly price back-calculated from 730 hours/month. On-demand and shorter terms are available by contacting sales.
| GPU | Configuration | Monthly (reserved) | Hourly |
|---|---|---|---|
| NVIDIA GH200 | 1 GPU, 96 GB GPU RAM, 72 vCPU, 480 GB RAM, 4800 GB storage, 25 TB bandwidth | $2,913 | $4.335 |
| NVIDIA H100 (8 GPU node) | 8 GPUs, 640 GB GPU RAM, 224 vCPU, 2048 GB RAM, 32640 GB storage, 15 TB bandwidth | $13,432 | $19.988 |
| AMD MI300X | Starting at | — | $2.000/GPU/hr |
| AMD MI325X | Starting at | — | $2.000/GPU/hr |
| AMD MI300X (fractional) | Starting at | — | $3.500/GPU/hr (full) / $2.290/GPU/hr (fractional) |
For on-demand and preemptible AMD Instinct pricing, Vultr also lists lower rates on spot-style instances — useful for batch AI training jobs that can tolerate interruption.
Beyond VMs and GPUs, Vultr's catalog extends into the rest of the cloud stack. Pricing on these is less grid-friendly but worth knowing.
Single-tenant dedicated servers. Vultr announced a $185/month entry bare metal plan (6 cores / 12 threads / 32 GB RAM / NVMe) for budget-conscious dedicated workloads, while on-demand bare metal pricing on the main pricing page starts around $750/month ($1.042/hr) and scales up past $1,000/month ($1.389/hr) depending on configuration. For multi-GPU bare metal and long-term contracts, you'll need to contact sales.
VKE's headline selling point is that the managed control plane is free — unlike AWS EKS or GKE, which typically charge $70+/month per cluster just for the management layer. You only pay for the underlying worker nodes (billed at standard Cloud Compute rates), plus any Load Balancers ($10/month) and Block Storage you attach. For a team running several small clusters, that's a meaningful savings line item.
👉 Deploy with Vultr Kubernetes Engine
S3-compatible, with two tiers:
- Standard Tier: $18/month base + $0.018/GB for additional storage
- Premium Tier: $36/month base + $0.018/GB for additional storage
A 2025 price increase roughly doubled object storage costs, which drew some community criticism — Backblaze B2 and S3 Infrequent Access remain cheaper for pure archival use.
NVMe SSD or HDD attachable storage starting at $1.00/month per 10 GB, scaling up to 40 TB HDD or 10 TB NVMe SSD per volume.
Fully managed MySQL, PostgreSQL, and Redis clusters starting at $15/month. Pricing scales with the underlying compute instance size and replication factor.
👉 Provision a Managed Database
Vultr's sticker prices are honest, but the add-ons are where bills drift upward. Based on what's confirmed on the official docs and FAQ:
- Bandwidth overage: $0.01/GB beyond the per-plan allowance and the 2 TB/month account-wide pooled egress
- Automatic backups: 20% of the instance price (so a $10/month instance costs $2/month for backups); the docs list $0.050/GB in some places
- Stored snapshots: $0.05/GB per month
- Reserved IP addresses: $0.005/hour when not attached to a running instance
- Load Balancers: $10/month base
- Object Storage egress: Variable by region, but generally counted against bandwidth pool
A Reddit user who switched from DigitalOcean to Vultr specifically noted backup pricing as the savings driver — Vultr's flat 20% surcharge beats DigitalOcean's 20%-of-droplet model when your instance costs more than $6/month. The flip side: if you snapshot aggressively, those $0.05/GB charges can quietly add up across dozens of servers.
Vultr runs multiple welcome offers in parallel. New accounts can typically pick one of the following (they cannot be combined):
- $200 free credit — promo code
FLYTWOHUNDRED - $250 free credit — promo code
250VULTRFLY - $300 free credit — applied via the signup funnel
- VULTRMATCH — Vultr matches your first deposit dollar-for-dollar up to $100 (so $100 in = $200 total credit)
All promotional credit is valid for 30 days and applies to select products only. The $300 offer tends to surface most often via affiliate signup links; the deposit-match is best if you already know you'll be spending real money quickly.
👉 Claim Vultr free credits and start deploying
Beyond credit promos, Vultr runs a genuine Free Tier Program for eligible developers. The free instance specs:
- 1 vCPU
- 512 MB RAM
- 10 GB SSD
- IPv4 address included
- 2 TB of free global bandwidth
The catch is availability — the program is currently limited to Miami, Seattle, and Frankfurt data centers, with limited quantities rolled out via a weighted-score application system. Anyone can apply, but admission is randomized with a weight given to how much identity and use-case information you provide. If admitted, the free instance appears on the deploy page under Cloud Compute > Regular > Free Instance.
👉 Apply for the Vultr Free Tier
Putting Vultr's numbers next to the usual suspects helps frame whether it's actually cheap. Based on competitor comparisons (VPSBenchmarks, Fluence's published comparison, Reddit migration threads):
| Workload | Vultr | DigitalOcean | Linode/Akamai | AWS Lightsail |
|---|---|---|---|---|
| 1 vCPU / 1 GB / 25 GB | $5–$6/mo | $6/mo | $5/mo | $5/mo |
| 2 vCPU / 4 GB / 80 GB | $20–$24/mo | $24/mo | $24/mo | $20/mo |
| 4 vCPU / 8 GB / 160 GB | $40–$48/mo | $48/mo | $48/mo | $40/mo |
| Kubernetes control plane | Free | Free (DOKS) | Free (LKE) | N/A |
| Bandwidth overage | $0.01/GB | $0.01/GB | $0.01/GB | varies |
Vultr lands roughly on par with DigitalOcean and Linode at the entry level and pulls ahead at the dedicated-CPU tier thanks to the VX1 line's price-performance improvements. Where Vultr actually differentiates is GPU pricing (H100 and MI300X at rates competitive with specialist GPU clouds), free Kubernetes control plane, and the breadth of 33 regions — more than DigitalOcean's 14 and Linode's 11.
A separate cost-per-dollar analysis by Fluence (a competing decentralized compute provider) argued that Vultr's effective price climbs once egress fees and bandwidth overages are factored in, particularly for distributed workloads with high outbound traffic. That's a fair point for data-egress-heavy apps, but for typical web hosting and dev workloads that stay well within the 2 TB pooled allowance, the comparison is mostly a wash.
VPSBenchmarks has tested 10 Vultr plans as of mid-2026, and the results paint a more nuanced picture than the marketing. Grades by plan (A best, F worst):
| Plan | Web Perf | Raw CPU | Stability | Disk IO | Network |
|---|---|---|---|---|---|
| High Frequency 1GB ($6) | D | F | D | C | D |
| Regular 2GB 1 core ($10) | E | F | C | C | D |
| High Performance 2GB ($12) | D | F | D | D | E |
| High Performance 4GB ($24) | D | D | C | C | C |
| CPU Optimized 2-4-50 ($40) | D | E | C | C | C |
| High Performance 8GB ($48) | D | C | B | C | C |
| VX1 GP 2C 8G 120S ($55) | B | D | A | A | C |
| VX1 GP 4C 16G 240S ($110) | B | C | B | B | B |
The takeaway: the cheapest shared-CPU plans grade poorly on raw CPU (they're throttled under sustained load), while the newer VX1 Optimized Cloud Compute plans score at the top of the chart — B and A grades across web performance, stability, and disk IO. If you're running anything more demanding than a personal blog, the jump from a $12 shared plan to a $55 VX1 plan is where the actual value lives. Vultr's average provisioning time across tested instances was 85 seconds, and its consistency score landed at 61 — solid mid-pack.
Matching plans to workloads is more useful than chasing the lowest sticker price. Here's how the tiers shake out in practice:
Personal blog or static site: Regular Performance 1 vCPU / 1 GB / 25 GB at $5/month is enough. Skip the $2.50 IPv6-only plan unless your visitors are all on IPv6 — many ISPs still aren't.
WordPress or low-traffic CMS: High Frequency 1 vCPU / 1 GB / 32 GB at $6/month. The 3 GHz+ CPU and NVMe storage make the back-end noticeably snappier than Regular Performance for the same dollar.
Small SaaS / API server: High Performance 2 vCPU / 4 GB / 100 GB at $24/month. NVMe SSD + AMD EPYC handles concurrent requests without choking.
E-commerce or production web app: General Purpose 2 vCPU / 8 GB / 50 GB at $60/month (VX1). Dedicated vCPU means no noisy-neighbor surprises during traffic spikes.
In-memory cache or analytics: Memory Optimized 2 vCPU / 16 GB / 200 GB at $100/month. The RAM-to-CPU ratio is what matters here.
Heavy database / NoSQL: Storage Optimized 4 vCPU / 32 GB / 960 GB at $310/month. The NVMe capacity lets you keep the hot dataset on fast local storage.
CI/CD or video encoding: CPU Optimized 4 vCPU / 8 GB / 150 GB at $90/month. Cores over RAM, exactly what batch jobs want.
AI inference / small model training: Start with a fractional AMD MI300X at $2.29/GPU/hour to prototype, then move to a reserved H100 node once you have stable demand.
Multi-cluster production Kubernetes: VKE with General Purpose worker nodes. Free control plane plus per-second-billed workers makes multi-cluster cost predictable.
Vultr cloud pricing is genuinely competitive once you understand the structure, but the structure does take a minute to learn. The headline $2.50/month entry point is real, the $6 High Frequency tier is where most small-project owners should actually start, and the VX1 Optimized Cloud Compute line is where the price-performance story actually delivers on its marketing claims. The free Kubernetes control plane, the 33-region footprint, and the aggressive GPU pricing are the differentiators worth pulling on if you're comparing Vultr against DigitalOcean, Linode, or even the hyperscalers for a new project.
The biggest pricing traps to watch for are bandwidth overages on data-egress-heavy workloads, snapshot costs if you're snapshotting liberally, and the gap between on-demand and reserved GPU pricing — which can be 2-3x depending on contract length. If you're a new account, the $300 free credit and the deposit match are both worth claiming before you commit to a real monthly spend, and the Free Tier Program is worth an application if you're an individual developer who can wait for the randomized admission.
Start with a free credit, deploy the smallest plan that covers your workload, and scale up only when the benchmarks (or your own monitoring) say you need to. That's the boring, correct way to use Vultr cloud pricing to your advantage.