Your cookie choices

We use necessary cookies to make this site work. With your permission, analytics cookies help us understand usage and marketing cookies support advertising.

You can reject non-essential cookies and still use the site. Change your choices anytime in Cookie settings.

Cookie policy
Cookie settings

Salad review

Updated September 28, 2026 · DeployCue Editorial Team · Facts checked September 26, 2026

We earn a commission when you sign up through links on this site. Partner status can affect where a brand appears in our rankings.

Our verdict

Salad is a distributed GPU cloud for developers and small teams who want cheap, interruptible single-GPU compute rather than fixed hardware commitments. GPU hours cost $0.015 and billing is per second, which suits model serving, rendering and batch work you can stop and restart.

Key facts

Best for
Burst single-GPU inference and batch work priced at $0.015 an hour.
GPU models
NVIDIA RTX 3060, NVIDIA RTX 3070, NVIDIA RTX 3070 Ti, NVIDIA RTX 3090, NVIDIA RTX 4090, NVIDIA RTX 5080, NVIDIA RTX 5090, NVIDIA RTX PRO 6000 Blackwell, Selected AMD classes (Source, checked September 26, 2026)
GPU hour from
$0.015 (Source, checked September 26, 2026)
Billing granularity
Per second (Source, checked September 26, 2026)
Locations
from 190 (Source, checked September 26, 2026)
Interconnect
None - single-GPU nodes, no multi-GPU NVLink interconnect (Source, checked September 26, 2026)
Compliance certifications
SOC 2 Type I (Source, checked September 26, 2026)
Legal entity
Salad Technologies, Inc. (Source, checked September 26, 2026)
Headquarters
United States (Source, checked September 26, 2026)
Founded
2018 (Source, checked September 26, 2026)
Support channels
Email, Help-centre, Community (Source, checked September 26, 2026)
Pricing model
Hourly billing
Official website
salad.com

LLM gateways

Fee model
Flat monthly subscription or pay per token (Source, checked September 26, 2026)
Bring your own key
No (Source, checked September 26, 2026)
OpenAI-compatible API
Yes (Source, checked September 26, 2026)
Free tier
No (Source, checked September 26, 2026)
Compliance certifications
SOC 2 Type I (Source, checked September 26, 2026)
Legal entity
Salad Technologies, Inc. (Source, checked September 26, 2026)
Headquarters
United States (Source, checked September 26, 2026)
Founded
2018 (Source, checked September 26, 2026)
Support channels
Email, Help-centre, Community (Source, checked September 26, 2026)

Kubernetes

Node price from
$0.015 (Source, checked September 26, 2026)
Locations
from 190 (Source, checked September 26, 2026)
Compliance certifications
SOC 2 Type I (Source, checked September 26, 2026)
Legal entity
Salad Technologies, Inc. (Source, checked September 26, 2026)
Headquarters
United States (Source, checked September 26, 2026)
Founded
2018 (Source, checked September 26, 2026)
Support channels
Email, Help-centre, Community (Source, checked September 26, 2026)

Serverless

Compliance certifications
SOC 2 Type I (Source, checked September 26, 2026)
Legal entity
Salad Technologies, Inc. (Source, checked September 26, 2026)
Headquarters
United States (Source, checked September 26, 2026)
Founded
2018 (Source, checked September 26, 2026)
Support channels
Email, Help-centre, Community (Source, checked September 26, 2026)

Apps and profiles

Elsewhere
LinkedInXgithub.comyoutube.comreddit.com

Pros and cons

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.
Salad

Ready to try Salad?

Scored 9.6 / 10 by DeployCue.

See alternatives

Who it suits

  • Developers running inference, rendering or batch jobs on consumer NVIDIA cards who can restart work after an interruption.
  • Small teams that want capacity spread thin across many locations rather than concentrated in a few regions.
  • Teams with an existing OpenAI-compatible client that can be pointed at the Salad AI gateway without a rewrite.

Who should look elsewhere

  • Multi-GPU training. Nodes are single-GPU with no NVLink (None - single-GPU nodes, no multi-GPU NVLink interconnect), while PoolCompute (NVIDIA Tensor Core H100 SXM, NVIDIA Tensor Core H100 NVL and NVIDIA Tensor Core H100 PCIe) and SberCloud (B300, H100 and A100) list parts built for datacentre work.
  • Buyers who want to hold their own model keys. Salad's gateway takes Flat monthly subscription or pay per token without bring your own key, and ComputePrices.com (H100 SXM, A100 SXM and H200) is a place to line up providers that publish wider catalogues.

Hardware and regions

Salad places container workloads across from 190 locations, so a job can sit near your users rather than inside a single fixed data center region. The card list is NVIDIA RTX 3060, NVIDIA RTX 3070 and NVIDIA RTX 3070 Ti, which leans on consumer and workstation NVIDIA parts rather than datacentre silicon, with selected AMD classes alongside. Interconnect is the constraint that shapes everything else: None - single-GPU nodes, no multi-GPU NVLink interconnect. Plan on one GPU per node, and scale a workload by adding nodes.

What it does well

Those hardware choices reward work that is parallel or restartable. The range in NVIDIA RTX 3060, NVIDIA RTX 3070 and NVIDIA RTX 3070 Ti covers image generation, video encoding and smaller model inference, with both older and current RTX generations plus a workstation-class part. A footprint of from 190 locations lets you spread replicas widely instead of concentrating them. The trade stays the same: None - single-GPU nodes, no multi-GPU NVLink interconnect, so throughput comes from many single-card nodes rather than one larger machine.

What it connects to

The Salad AI gateway exposes an OpenAI-compatible API, so SDK calls, client libraries and existing chat or completion code can be pointed at it with minimal change. Bring your own key is not offered, so provider credentials are not the ones on your side of the connection and the gateway bills through Salad.

Who runs it

The service is provided by Salad Technologies, Inc., headquartered in the US and founded in 2018, so contracts, invoicing and data handling sit with a US entity. Support runs on email, help centre and community, which leans on documentation and community answers rather than staffed live channels. That suits developers who are comfortable reading docs and comparing notes before opening a ticket.

How Salad compares

Frequently asked questions

What does a GPU hour cost on Salad?+

GPU hours cost $0.015, the lowest entry figure among the 95 providers ranked on that measure. The median across all listed brands is €0.63, so the opening gap is wide. Stronger cards and longer runs sit above that starting point.

How is compute billed?+

Compute is billed per second and the pricing model is listed as hourly. That combination suits bursty inference and batch work, where a job starts and stops without a long commitment. Container workloads share the same entry point, with node pricing of $0.015.

Is there a free tier to test with?+

A free tier is not offered, so a trial starts on a paid plan. Most providers in the GPU ranking do include one, and 27% of that ranking shares Salad's paid-only setup. Keep a first run per second and the test cost stays modest.

Which GPUs can I get?+

The list is NVIDIA RTX 3060, NVIDIA RTX 3070 and NVIDIA RTX 3070 Ti, so consumer and workstation NVIDIA parts dominate, with selected AMD classes as well. Compare that with PoolCompute's NVIDIA Tensor Core H100 SXM, NVIDIA Tensor Core H100 NVL and NVIDIA Tensor Core H100 PCIe or SberCloud's B300, H100 and A100 if a specific datacentre card is required. Pick by workload: older and current generations both appear.

Where does Salad actually run?+

Workloads sit in from 190 locations, far above the 6 median in the GPU ranking. A spread that wide suits jobs you want close to users, and it is the clearest difference between Salad and providers that concentrate capacity in a handful of regions.

Can I use several GPUs in one node?+

Not inside a single node. Interconnect is None - single-GPU nodes, no multi-GPU NVLink interconnect, so each job sees one card. Multi-GPU training plans belong with a provider listing datacentre silicon, such as PoolCompute (NVIDIA Tensor Core H100 SXM, NVIDIA Tensor Core H100 NVL and NVIDIA Tensor Core H100 PCIe) or SberCloud (B300, H100 and A100). On Salad, scale means more nodes, not wider ones.

Can I bring my own model key to the AI gateway?+

Bring your own key is not offered, and the fee model is Flat monthly subscription or pay per token. You do get an OpenAI-compatible API, so existing client code can point at the gateway as written, but the credentials in play stay on Salad's side. If owning the key matters to your setup, weigh that before you build on it.

What security certification does Salad list?+

Salad lists SOC 2 Type I in its security materials, and that is the attestation named for the service. If your procurement checklist calls for a particular report or scope, ask the company for it directly before you build a pipeline that depends on the gateway.

How do I reach support?+

Support runs on email, help centre and community. Email, a help centre and a community cover most questions, which suits developers who prefer written answers they can search later. No live chat or phone line is listed, so time-sensitive issues move faster over email to Salad Technologies, Inc..

Sources

  • pppreport.org/company/salad+technologies+inc checked September 26, 2026

    Backs: Headquarters

    “26 S Rio Grande St #2072 SALT LAKE CITY, UT 84101”

  • salad.com checked September 26, 2026

    Backs: Interconnect

    “Anything that needs sub-10-millisecond latency, multi-GPU NVLink training, or data-residency guarantees that a consumer network cannot provide.”

  • salad.com checked September 26, 2026

    Backs: Node price from

    “$0.015 per GPU-hour, starting price”

  • salad.com/about checked September 26, 2026

    Backs: Founded, Locations

    “2018 founded, March 27”

  • salad.com/ai-gateway checked September 26, 2026

    Backs: Bring your own key, Fee model, OpenAI-compatible API

    “Set OPENAI_BASE_URL=https://ai.salad.cloud/v1 and use your AI Gateway key as the API key.”

  • salad.com/download checked September 26, 2026

    Backs: Support channels

    “Help Portal (https://support.salad.com) | Discord (https://discord.gg/salad) | Reddit (https://www.reddit.com/r/SaladChefs/)”

  • salad.com/pricing checked September 26, 2026

    Backs: Billing granularity, Free tier, GPU hour from, Pricing model

    “Billing is per second while an instance is in the running state. It stops the moment the instance stops, is reallocated, or is preempted.”

  • salad.com/salad-container-engine-community checked September 26, 2026

    Backs: GPU models

    “RTX-class GPUs, RTX 3060 through RTX 5090”

  • salad.com/security checked September 26, 2026

    Backs: Compliance certifications

    “Salad Technologies is SOC 2 Type I compliant under AICPA standards and SSAE 18, audited by Prescient Assurance.”

  • salad.com/terms/saladcloud checked September 26, 2026

    Backs: Legal entity

    “operated by Salad Technologies, Inc. (the "Company")”

Compiled from 10 source pages, checked September 26, 2026. We did not open an account.

Not confirmed yet: Egress fee.

Salad

Salad

Burst single-GPU inference and batch work priced at $0.015 an hour.

9.6/10See alternatives