5 other AI Infrastructure & Deployment tools we have sourced reviews for, all in the same niche as Foundry (now Mithril) (GPU Cloud). Within that, they're ordered by aitrove popularity, not by a quality score: we haven't tested them hands-on, so we don't rank them as "best".
Renamed: Foundry rebranded as Mithril on August 4, 2025, at mithril.ai. Sources: [1]
Best for: Teams shopping for flexible GPU capacity.
Biggest catch: Rebranded from Foundry to Mithril in August 2025, so the old name and URL are stale.
Skip it if: You want a fully managed ML platform.
Price: Spot and reserved pricing on mithril.ai (varies by GPU). (checked Oct 11, 2026)
| Tool | Best for | Biggest catch | Starting price |
|---|---|---|---|
| Paperspace | Teams already on DigitalOcean who need GPU VMs. | Legacy Gradient and Core APIs are deprecated and unavailable | Pay-as-you-go GPU pricing via DigitalOcean (check its pricing page). (checked Oct 11, 2026) |
| Lambda | ML teams who want straightforward GPU rentals with no egress fees. | H100 on demand is $3.99/GPU-hr, higher than marketplace-style clouds | H100 SXM from $3.99/GPU-hr (8x), H100 PCIe $3.29/GPU-hr. Clusters from $5.54/GPU-hr. (checked Oct 11, 2026) |
| RunPod | Hobbyists and startups doing disposable GPU work. | Community Cloud is peer-to-peer, with variable reliability and no SLA | Pay-as-you-go. E.g. H100 PCIe about $1.99/hr Community, $2.89/hr Secure (third-party report). (checked Oct 11, 2026) |
| Together AI | Developers serving or fine-tuning open models. | Serverless rates vary by model, so you need to check each one | Serverless per million tokens (by model). H100 clusters $3.99/GPU-hr on demand, $1.99 preemptible. (checked Oct 11, 2026) |
| Vast.ai | Budget experiments and fine-tuning jobs that tolerate interruptions. | Reliability depends on the host, and interruptions and storage loss have been reported | Marketplace pricing. H100 roughly $0.90–$1.87/hr (third-party reports). (checked Oct 11, 2026) |
Best for: Teams already on DigitalOcean who need GPU VMs.
Legacy Gradient and Core APIs are deprecated and unavailable
Skip it if: You relied on the old Gradient workflows.
Starting price: Pay-as-you-go GPU pricing via DigitalOcean (check its pricing page). (checked Oct 11, 2026)
Best for: ML teams who want straightforward GPU rentals with no egress fees.
H100 on demand is $3.99/GPU-hr, higher than marketplace-style clouds
Skip it if: You need the cheapest possible spot GPUs.
Starting price: H100 SXM from $3.99/GPU-hr (8x), H100 PCIe $3.29/GPU-hr. Clusters from $5.54/GPU-hr. (checked Oct 11, 2026)
Best for: Hobbyists and startups doing disposable GPU work.
Community Cloud is peer-to-peer, with variable reliability and no SLA
Skip it if: You need guaranteed production reliability on the cheap tier.
Starting price: Pay-as-you-go. E.g. H100 PCIe about $1.99/hr Community, $2.89/hr Secure (third-party report). (checked Oct 11, 2026)
Best for: Developers serving or fine-tuning open models.
Serverless rates vary by model, so you need to check each one
Skip it if: You want a consumer chat app.
Starting price: Serverless per million tokens (by model). H100 clusters $3.99/GPU-hr on demand, $1.99 preemptible. (checked Oct 11, 2026)
Best for: Budget experiments and fine-tuning jobs that tolerate interruptions.
Reliability depends on the host, and interruptions and storage loss have been reported
Skip it if: You run production or regulated workloads.
Starting price: Marketplace pricing. H100 roughly $0.90–$1.87/hr (third-party reports). (checked Oct 11, 2026)
Pros, catches and prices are compiled from public sources (official pricing and docs pages, published reviews, user reports), not hands-on testing. Prices change often, so check the vendor page before you buy.