Top serverless and edge compute compared

About Top serverless and edge compute compared

Serverless and edge compute pricing rests on a meter for run time and a meter for requests. This ranking covers 176 brands in the category, drawn from 523 providers tracked on the site, and it places the ones that publish both kinds of price near the front. The trade-off readers settle here runs through price at entry, price at renewal, what the plan includes and where the workload can run. SberCloud prices run time and requests in rubles, adds a free allowance of 1,000,000 requests, and answers across email, phone and chat. Google Cloud prices the same pair in dollars with a free allowance of 2,000,000 requests, but its help runs on help centre and community. RunPod stretches duration furthest, with endpoints configured from 5 to 604,800, and answers on email, phone and help centre. Further down the list the comparison rests on the contracting entity, the headquarters country and the listed support channels, because those are the details a reader can check before an early deployment. Introductory pricing is read apart from renewal pricing throughout, so a low opening figure never stands in for the ongoing rate.

How we picked

Scores here combine weighted inputs. Price carries 40%, what the plan includes carries 40%, and trust carries 20%. Price covers entry pricing, compute rates per GB-second of run time and per-request charges, with the request rate for this ranking sitting at €0.35 and introductory terms read apart from renewal terms. Offer covers free request allowances, maximum request duration, billing granularity and how far a workload spreads across locations. Trust covers the contracting legal entity, headquarters country, founding year and the support channels a provider answers on. Sources are published vendor pages: pricing pages, tariff documents, terms of service, support and contact pages, and encyclopedia entries for founding dates and headquarters. Site-wide and category-wide figures come from the site database of 523 providers, with 176 sitting in this ranking and 176 carried in the page data. Ordering is by score. Partner entries sit ahead of scored entries, and every entry here sits on score alone, with the labels shown as computed. Where providers land close together, the blurb for each names the fact that separates it from the member ranked next to it, and every con points at the fact it rests on.

Our top picks

1. 2. RunPod

   RunPod closes this ranking on the facts that matter for long jobs: a duration ceiling of 5 to 604,800 per endpoint and support across email, phone and help centre, covering email, a phone line, a help centre and a community. It is a US company founded in 2022 and contracting as Runpod Inc.. The trade-off against Google Cloud above is the meter: Google Cloud caps a request at 3,600, while RunPod spans a far longer window, which suits batch inference and background jobs more than burst handlers.

   • GPU models: B300, H200, B200, and 18 more
   • GPU hour from: $0.16
   • Billing granularity: Per second

   Pros

   • GPU hours start at $0.16, under the median entry rate of €0.63 across providers listed here.
   • Moving data in or out adds nothing to a bill: No charge - no fees for data egress or ingress.
   • Billing is per second, so tests and short jobs stop paying for time they do not use.
   • The catalogue holds current datacentre parts alongside older consumer and workstation cards: B300, H200 and B200.
   • Support spans email, phone and help centre, including phone for the cases where a job is stuck.

   Cons

   • ComputePrices.com lists families outside this catalogue, H100 SXM, A100 SXM and H200, including V100 and AMD parts.
   • The only compliance certification named is GDPR, so regulated buyers have a narrow set to point at.
   • Serverless endpoint configurations are limited to 5 to 604,800, which rules out long batch jobs on that path.
   • The listed billing model is hourly, so budgets have to be built in hours rather than per request.

2. 5. SberCloud: best for top-pick

   SberCloud earns the top-pick label in this ranking because it publishes a complete serverless price sheet. The FunctionGraph service bills on a run-time meter and a request meter: a per-GB-second rate of RUB 0.0013 and a request rate of RUB 14.34, plus a free allowance of 1,000,000 requests. Support runs across email, phone and chat, the widest mix in this ranking, which matters when a function fails outside working hours. The trade-off sits with location and contract. The operation is headquartered in RU and signs through a Russian limited liability company, so teams billing in other currencies carry questions that Salad below does not raise.

   • Price per GB per month: RUB 814.68
   • Egress price per GB: RUB 1.15
   • Free egress: 10 TB of outgoing traffic per month

   Pros

   • Outgoing traffic includes 10 TB of outgoing traffic per month each month, so routine transfers cost nothing before RUB 1.15 per GB applies.
   • A free tier plus bonus credits for individuals, legal entities and sole proprietors keeps early testing cheap.
   • Evolution compute starts at RUB 23.75 with 1 vCPU, 1 GB RAM, 30 GB NVMe, billed hourly.
   • GPU capacity covers B300, H100, A100, V100 and A40 from RUB 240 an hour, billed per minute.

   Cons

   • Kubernetes has no free control plane, so every cluster adds RUB 2.98 on top of the node price.
   • Infrastructure sits in RU across 9 locations, so there is no reach outside that market.
   • Storage covers object only, so block and file tiers are not part of the same plan.

3. 6. Nosana

   Nosana is the only member of this ranking based in the Netherlands, with a legal entity of Nosana BV and a founding year of 2021. Support runs on email, help centre and community, email, a help centre and a community, and no phone line appears in that list, a thinner channel set than Vultr below, which also lists no phone and no email. For readers in this ranking, Nosana is the pick when an EU contracting entity matters and written documentation matters more than voice support, while Vultr below wins on age, having been founded in 2014.

   • GPU models: NVIDIA 3060, NVIDIA 4060, NVIDIA 3070, and 18 more
   • GPU hour from: $0.048
   • Free credits: Free credits when signing up with email

   Pros

   • The card list in NVIDIA 3060, NVIDIA 4060 and NVIDIA 3070 covers consumer, workstation and datacenter parts, including the A100 and the H100.
   • A GPU hour from $0.048 puts entry well under the ranked median of €0.61.
   • Signing up with an email address brings free credits, which lowers the cost of a first test run.

   Cons

   • Billing is limited to hourly, so there is no reserved or monthly rate to lock in for steady training.
   • Support runs on email, help centre and community only, with no phone or live chat path listed.

4. 7. Salad

   Salad ranks directly behind SberCloud in this ranking on the strength of a US base and a settled contracting entity. Founded in 2018, the company signs as Salad Technologies, Inc., so the paperwork side of an early deployment is plain. Support runs on email, help centre and community, a help centre and email, which is a thinner mix than the one above but enough for teams that work from documentation and tickets. Readers weighing this ranking should hold the support question against price at entry and again at renewal, and PoolCompute below answers the same questions with a US operation formed in 2024 and a narrower channel set.

   • GPU models: NVIDIA RTX 3060, NVIDIA RTX 3070, NVIDIA RTX 3070 Ti, and 6 more
   • GPU hour from: $0.015
   • Billing granularity: Per second

   Pros

   • Billing is per second against an hourly model, so short runs and interrupted jobs pay only for the time they hold a card.
   • Workloads sit in from 190 locations, well above the 6 median in the GPU ranking.
   • The catalogue in NVIDIA RTX 3060, NVIDIA RTX 3070 and NVIDIA RTX 3070 Ti spans several RTX generations, the RTX PRO workstation class and selected AMD classes.
   • The AI gateway is OpenAI-compatible, so clients already written against that API can be repointed at Salad.
   • The operator is Salad Technologies, Inc., a US company founded in 2018, with SOC 2 Type I listed for security.

   Cons

   • One GPU per node: None - single-GPU nodes, no multi-GPU NVLink interconnect, so any plan that needs several accelerators inside a single machine has to look elsewhere.
   • Bring your own key is not offered, and the gateway is billed as Flat monthly subscription or pay per token, so you cannot pay through your own provider account.
   • Support runs on email, help centre and community only, with no live chat or phone line listed for urgent issues.
   • A free tier is not offered, and only 27% of the GPU ranking sits in the same paid-only position.

5. 10. Google Cloud

   Google Cloud is in this ranking because it is the only member publishing its meters in dollars, which turns the comparison with SberCloud into a currency exercise. Cloud Run charges a request rate of $0.40 and a run-time rate of $0.000002 to $0.0000025, with a free allowance of 2,000,000 requests. A request is capped at 3,600, and support runs on help centre and community, a help centre and a community forum with no email and no phone listed. The GB-second rate reads as a range, so the tier sets the figure, while Supascale AI below runs a newer operation with a thinner channel mix.

   • Control plane price: $0.10
   • Free control plane: Yes
   • HA control plane price: $0.10

   Pros

   • The control plane is free, and the paid tier sits at $0.10, close to the site median of €0.06.
   • Scale headroom is large: clusters run to 65,000 nodes, well past what most managed offerings in the ranking allow.
   • A free tier and $300 in free credits in credits cover early experiments before any spend.
   • Egress includes 100 a month for North America transfers, which cuts the $0.12 per-GB rate for small-scale traffic.
   • Footprint is wide, with 40 locations and 46 regions, so workloads can sit close to users.

   Cons

   • The control plane is free only inside the allowance, so sustained clusters bill at $0.10 for the standard tier and $0.10 for the highly available one.
   • Egress is charged at $0.12 per GB once the free allowance is used, and traffic to Australia and China is excluded from it.
   • Support runs on help centre and community only, so paid and phone channels are not part of the standard offering.

What to look for

Entry price against renewal price

Start with what the opening invoice looks like, then read the renewal rate beside it. The median entry price in this ranking is €3.68, and most sit €2.43 to €7.42. A provider whose opening rate is well under the median can still renew far above it.

Free request allowance

The allowance decides how a low-traffic service behaves before any charge lands. The median across this ranking is 1,000,000 free requests, with 10,000,000 at the top of the range and 1,000 at the bottom. Anything near the bottom serves hobby work only.

Request fee, compute fee and egress

A request fee and a GB-second fee reward different shapes of workload. The median request rate in this ranking is €0.35, and egress adds a line of its own at €0.01 per GB. Chatty endpoints and compute-heavy ones land in different places on that meter.

Support that answers

Where a provider answers matters as much as what it charges. A help centre covers 81.3% of this ranking, a community forum covers 76.6%, email covers 69.6% and a phone line covers 21.1%. The thinner the mix, the more the plan depends on written answers.

Where the workload runs

Footprint decides latency and data residency. The median location count in this ranking is 15, while the median count of points of presence is 2,600. A single-region provider suits regulated traffic, and a wide spread suits users spread across continents.

Billing granularity

Short functions only stay cheap when the meter is fine. Per-second billing covers 48.9% of this ranking, per-minute covers 24.4% and hourly covers 26.7%. Bursty event handlers reward the fine end of that list.

Frequently asked questions

Top serverless and edge compute compared by country

• Brazil
• Canada
• China
• France
• Germany
• India
• Indonesia
• Italy
• Japan
• Mexico
• South Korea
• Spain
• United Kingdom
• United States
• Vietnam

Providers compared head to head

• Coconut vs Codemagic
• Coconut vs CodeSandbox
• Coconut vs Codesphere
• Codemagic vs CodeSandbox
• Codemagic vs Codesphere
• CodeSandbox vs Codesphere