Shadeform

shadeform.ai · Compute · 1 record · JSON

Shadeform Compute

paidemail?card?Drequest-rate

www.shadeform.ai/ · GPU cloud · compute/shadeform · JSON · compare

Free tier	not stated unit none; metered on wall-clock
Price	none stated
Per-object cap	not stated
Retention	not stated
Egress	not stated
Identity	email (placeholder) not published
Limits	request-rate GPU instances across 30+ providers with centralised billing; reserved capacity carries a 12-month minimum term source per-ip-shaping the marketplace claims savings of up to 90% against AWS, GCP and Azure source not stated
Binding	request-rate
Interface	unknown
Tooling	not stated
Composes with	not stated
Reliability	no reliability event recorded
Evidence rung	documented the site was read on 2026-10-22 and describes a GPU marketplace with centralised billing and a 12-month minimum term on reservations. No free tier appears, which confirms paid-only
Checked	2026-10-23 (live); stale on 2026-11-22

Evidence (15)

date	source	what it proved
2026-10-02	www.shadeform.ai/	not published
2026-10-22	www.shadeform.ai/	'The GPU Cloud Marketplace · Deploy high quality GPUs at the best prices across 30+ top clouds worldwide' with 'one console to provision GPU instances across top providers with centralized billing' and 'Minimum term is 12 months (1 year)', quoted
2026-10-22	shadeform.com/prices	'No fees, no markups.' as the page heading, with 'Using Shadeform is free. You pay the same as going direct to the clouds in our network.', quoted. The page publishes no rate table at all, so every rate in this finding comes from the instance directory, not from the pricing page
2026-10-22	docs.shadeform.ai/getting-started/faq.md	'Shadeform charges the same rate as going direct to the cloud provider. There is no additional fee for using Shadeform to launch instances or manage cloud accounts.' and 'When you launch an instance through Shadeform, you pay Shadeform for the GPU time and Shadeform passes the payment to the underlying cloud provider.', quoted from the Business section
2026-10-22	docs.shadeform.ai/getting-started/faq.md	Shade Cloud is 'the name of the feature where we deploy your instances into our network of cloud accounts with each of the cloud providers. By using Shade Cloud, you can use the underlying cloud provider without having an account with them.' and 'As Shadeform is an abstraction for different clouds and data centers, we inherit the SLAs provided by cloud providers.', quoted
2026-10-22	shadeform.com/directory/instances/hyperstack/A4000	'A4000 \| $0.15/hr' with 'pcie\|16 GB VRAM\|VM\|0 available regions', vCPUs 4, Memory 21 GB, Storage 100 GB; the on-demand table on the same page lists 'A4000 \| Hyperstack \| $0.15 /hr' beside 'A4000 \| Paperspace \| $0.80 /hr', quoted. The unit is the whole VM, not a GPU attach
2026-10-22	shadeform.com/directory/instances	sorting the directory by the Price column ascending put 'Hyperstack A4000 $0.15 - 1 16 No regions PCIe 21 100 4' first, then 'Hyperstack A4000 $0.30', 'Massed Compute A30 $0.35', 'Verda V100 $0.39', 'Massed Compute A5000 $0.44', 'IMWT A6000 $0.49', 'Hyperstack A6000 $0.50', 'Vultr A16 $0.51' and 'Massed Compute A6000 $0.57' twice, on Page 1 of 28; the descending sort started at 'Nebius B300 $88.00 $56.00 8 2304'
2026-10-22	www.hyperstack.cloud/gpu-pricing	the host's own on-demand rows 'NVIDIA A4000 16 6 24 $0.15', 'NVIDIA A6000 48 28 58 $0.50' and 'NVIDIA H100 80 28 180 $2.50', against Shadeform's Hyperstack rows at USD 0.15, USD 0.50 and USD 2.50: exact equality on three cards, which supports the no-markup claim for this host
2026-10-22	www.paperspace.com/pricing	the host's hourly block reads 'A4000 $0.76 / hour' with 'NVIDIA A4000 GPU, 45GB RAM, 8 vCPU', and its monthly block on the same page reads 'A4000 $488 / month' with 44GB RAM and 8 vCPU, so USD 0.76 is the hourly rate and not a committed one; Shadeform lists the same 8 vCPU and 45 GB machine at USD 0.80/hr, a +USD 0.04 (+5.3%) gap
2026-10-22	docs.shadeform.ai/getting-started/quickstart.md	'After you verify your email address, a two-step form asks for your name and organization (your name appears on invoices and team invites; the organization names your workspace), then a few questions about what you plan to run.' and 'Add Funds makes a one-time payment through Stripe (minimum $20)', quoted
2026-10-22	docs.shadeform.ai/getting-started/quickstart.md	'Instances are billed per second against your account balance' and, of a launch with no funds, 'the console asks you to add funds before it continues', quoted; the page sets Deploy in bold inside that sentence
2026-10-22	docs.shadeform.ai/getting-started/concepts.md	Shade Cloud 'deploys the instance into its own account with the underlying cloud provider', and a Flex VM is 'a GPU instance that Shadeform virtualizes itself on a partner's bare-metal hardware, rather than one the provider virtualizes in their own cloud', quoted: the virtualisation layer is Shadeform's, the iron is a partner's
2026-10-22	docs.shadeform.ai/guides/spotinstances.md	'Shadeform may reclaim the capacity at any time, at which point the instance is terminated', quoted; every entry of an instance type carries a rental_type, and a spot entry carries its own hourly price
2026-10-22	www.shadeform.ai/	'Deploy high quality GPUs at the best prices across 30+ top clouds worldwide.' with 'Shadeform supports over 100 regions across North America, Europe, Asia and Australia.' and, further down the same page, 'the best prices for your workloads across 150+ regions': one page carries two different region counts
2026-10-22	shadeform.com/directory/clouds	17 clouds named across Page 1 of 2 and Page 2 of 2: Boost Run, Crusoe, Cudo Compute, Denvr Dataworks, DigitalOcean, Phyntec, Gcore, Hyperstack, Lambda, Latitude, Massed Compute, Nebius, Paperspace, Scaleway, Verda Cloud, Voltage Park and Vultr; the instance table adds IMWT and Excess Supply, which the cloud directory does not list

More fields (16)

{
 "talaria_m10h_usd": null,
 "talaria_m24h_usd": null,
 "nemo_m10h_usd": null,
 "keep_rate": null,
 "keep_rate_basis": "no field in the source",
 "billing_model": "unclassified",
 "shape": null,
 "session_cap": {
  "wall_clock_min": null,
  "idle_reap_s": null,
  "restart_required": false,
  "note": "on-demand instances across providers; reserved capacity has a 12-month floor"
 },
 "persistent_storage": null,
 "network_out": null,
 "price_note": null,
 "free_tier_kind": "none",
 "reclassified_from": "dead",
 "ps": null,
 "net": {
  "out": true,
  "in": true,
  "ports": null,
  "static_ip": null,
  "ipv6": null,
  "allowance": null
 },
 "wave2": {
  "merged_agents": [
   "shadeform"
  ],
  "merged_evidence_rows": 13,
  "merged_note": "A GPU broker. The console, REST API, CLI and MCP server list instances from 19 clouds named on the vendor's own directory pages, and the cheapest row in the whole directory on 2026-10-22 is an NVIDIA A4000 16 GB on Hyperstack at USD 0.15 per hour for a 1-GPU VM with 4 vCPU, 21 GB RAM and 100 GB storage. Next rows: A4000 x2 USD 0.30, Massed Compute A30 24 GB USD 0.35, Verda V100 16 GB USD 0.39, Massed Compute A5000 USD 0.44, IMWT A6000 USD 0.49, Hyperstack A6000 USD 0.50, Vultr A16 USD 0.51, Massed Compute A6000 USD 0.57. Top of the same table: Nebius B300 8x USD 88.00 (spot USD 56.00), Nebius "
 }
}

Unknown or withheld values (19)

$.chaining not-stated
$.egress.metered egress-unknown
$.egress.terms not-stated
$.evidence[0].proved process
$.extras.price_note process
$.free_tier.notes withheld
$.free_tier.stated not-stated
$.identity_cost.detail defaulted
$.identity_cost.kind identity-defaulted
$.interface interface-unknown
$.interface_detail withheld
$.limits_note not-stated
$.notes process
$.per_object_cap.notes process
$.per_object_cap.stated not-stated
$.reliability_note none-observed
$.retention.notes not-extracted
$.retention.stated not-stated
$.tooling not-stated