The most price-transparent provider of this set with published per-second rates, live availability and measured earn statistics from a clearly identified US company, but only consumer RTX-class GPUs are offered and workloads must tolerate preemption, with fiat-only buying.
by the bestcloudmining editorial desk · review version 2 · prose updated 2026-09-24 · facts as of 2026-09-24 · how we verify
SaladCloud is a US-based GPU marketplace offering distributed consumer-grade compute (RTX 30/40/50 series, AMD RX) for AI, rendering and batch workloads. It’s designed for developers needing affordable, interruptible GPU power with per-second billing and no KYC. While transparently priced and backed by a known operator, workloads must tolerate preemption since nodes run on users’ personal machines.
You can rent GPU compute from a pool of consumer-grade hardware including RTX 3060, 3090, 4090, 5090 and AMD RX 7900 XTX, accessed via Docker, API or Kubernetes. Each GPU includes bundled vCPU and RAM. There are no ASICs or datacenter-class hardware like H100s. You rent time, not ownership - your workload runs in isolated VMs on third-party machines that can disconnect at any time. You do not get physical access, hardware guarantees or uninterrupted runtime.
SaladCloud offers flexible, zero-day contracts with no long-term commitment - services are paid in advance via prepaid balance with optional auto top-up. There is no stated maintenance fee, but no uptime SLA either. Payouts for chefs occur within five business days as Reward Points, which are non-refundable and non-transferable for cash except via PayPal or approved redemption options. Refunds for unused prepaid balances are not detailed in the ToS.
There is no cryptographic proof-of-compute, TEE attestation or on-chain verification that your workload ran as intended. Workloads execute in isolated VMs on consumer hardware, and node reliability depends on individual users keeping their machines online. While Salad publishes live availability and measured earnings data (including idle time), actual execution integrity must be trusted - not verified. This is typical for consumer-distributed networks.
Salad Technologies, Inc. has operated since 2018, claims SOC 2 Type I compliance, and reports $7M+ paid to 450,000+ contributing GPUs across 190+ countries. External sources show consistent, real-world earnings data from long-term users that align with published rates. Red flags are minimal - no fraud allegations exist, though community discussions note common issues like support delays and payout processing. The operator is transparently backed and US-based, with clear legal terms.
Pricing starts at $0.015 per GPU/hour for entry-level cards (e.g., RTX 3060) at Lowest priority, billed per second while running. High-end options like the RTX 4090 range from $0.16 to $0.33/GPU-hr depending on priority. You pay in USD via Stripe with no KYC; there is no minimum purchase, but no free trial either. Prices are publicly listed and include vCPU and RAM - cold boot is free, and auto top-up can be configured.
SaladCloud is a legitimate, transparent option for developers needing low-cost, interruptible GPU power for batch or fault-tolerant workloads. It’s not suitable for mission-critical or low-latency tasks. If you need predictable uptime or enterprise-grade hardware, consider traditional cloud providers instead.
| Documented price | $0.015 per GPU/hr · $0.160 per GPU/hr (RTX 4090, Lowest priority) |
|---|---|
| Payouts | on redemption (Reward Points appear within 5 business days) |
| Notes | Published per-second billing, USD/GPU-hr by priority tier (High/Medium/Low/Lowest); vCPU and RAM included, cold boot free. Entry classes (RTX 3060/3070/3070 Ti etc.) from $0.015/GPU-hr at Lowest; published table RTX 5060 $0.065-0.11, RTX 4090 $0.16-0.33, RTX 5090 $0.25-0.50, RTX 3090 $0.09-0.17 (2026-09-24). No H100/A100. |
Run the long math before paying: cloud mining calculator →
Ready to try Salad / SaladCloud? Start small, verify one payout, and only scale up once the hashrate or compute actually matches what you paid for.
Based on our checks as of 2026-09-24: Salad / SaladCloud scores 7.5/10 in our weighted review. We logged no red flags. User reports and operator background are documented with dated sources in the meta-review below.
You get time on consumer GPUs (like RTX 30/40/50 series) in isolated VMs, with vCPU and RAM included. You do not own the hardware, and nodes can disconnect at any time - workloads must be fault-tolerant.
You pay only the published USD rate per GPU/hour, billed per second. There are no additional maintenance fees, but you must prepay via Stripe; no hidden charges are documented.
There is no stated minimum purchase, but no free trial is offered. You prepay into a balance and consume credits as you run workloads.
Contributors earn Reward Points based on usage, which appear within five business days. Points can be redeemed via PayPal, crypto (like RENDER), gift cards or charity - not as direct cash unless through approved methods.
If operations cease, your prepaid balance and active workloads would likely be lost. There’s no recourse for compensation, and redeemed rewards are non-refundable.
Renting avoids upfront hardware costs and maintenance but offers no asset ownership. Compared to buying coins, it’s a compute service - not an investment. It’s cost-effective for short-term AI/rendering jobs, but not for guaranteed returns.
Solana-based self-serve GPU marketplace for AI inference, CI and agent workloads; live price table, consumer-GPU supply.
Open permissionless marketplace for containerised GPU/CPU compute; providers bid, tenants pay per block.
Host-set GPU marketplace with per-second billing; on-demand, interruptible and reserved tiers, plus serverless and clusters.
No comments yet - be the first. Posting runs a short proof-of-work in your browser (anti-spam), no account needed.