Evidence-gated provider × GPU pages
Choose a provider and accelerator without losing the exact configuration evidence. Each page includes normalized rates, regional rows, availability boundaries and a workload budget calculator.
Canonical decision inventory
The inventory is rebuilt from official-source records. A page needs enough configurations, regional or provider diversity and distinct price points for its exact intent. It is not published merely because a keyword can be formed.
Start with Akamai Cloud Computing when provider choice is already constrained, then choose among the exact GPU models that currently pass the evidence gate.
$0.520–$1.530 per GPU-hour · 33 regions.
Deployability evidence: 297/297 rows (100%). This is not a live-stock claim.
160 GPU-hours: $83.20–$244.80 observed compute range.
730 GPU-hours: $379.60–$1116.90 observed compute range.
$1.500–$1.500 per GPU-hour · 33 regions.
Deployability evidence: 132/132 rows (100%). This is not a live-stock claim.
160 GPU-hours: $240.00–$240.00 observed compute range.
730 GPU-hours: $1095.00–$1095.00 observed compute range.
Start with Microsoft Azure when provider choice is already constrained, then choose among the exact GPU models that currently pass the evidence gate.
$0.679–$8.193 per GPU-hour · 22 regions.
Deployability evidence: 0/117 rows (0%). This is not a live-stock claim.
160 GPU-hours: $108.60–$1310.80 observed compute range.
730 GPU-hours: $495.50–$5980.53 observed compute range.
$1.290–$23.326 per GPU-hour · 31 regions.
Deployability evidence: 0/183 rows (0%). This is not a live-stock claim.
160 GPU-hours: $206.38–$3732.16 observed compute range.
730 GPU-hours: $941.63–$17027.98 observed compute range.
Start with Nebius AI Cloud when provider choice is already constrained, then choose among the exact GPU models that currently pass the evidence gate.
$3.950–$7.150 per GPU-hour · 2 regions.
Deployability evidence: 4/4 rows (100%). This is not a live-stock claim.
160 GPU-hours: $632.00–$1144.00 observed compute range.
730 GPU-hours: $2883.50–$5219.50 observed compute range.
$4.300–$7.850 per GPU-hour · 2 regions.
Deployability evidence: 4/4 rows (100%). This is not a live-stock claim.
160 GPU-hours: $688.00–$1256.00 observed compute range.
730 GPU-hours: $3139.00–$5730.50 observed compute range.
$2.450–$4.500 per GPU-hour · 4 regions.
Deployability evidence: 8/8 rows (100%). This is not a live-stock claim.
160 GPU-hours: $392.00–$720.00 observed compute range.
730 GPU-hours: $1788.50–$3285.00 observed compute range.
Start with OVHcloud when provider choice is already constrained, then choose among the exact GPU models that currently pass the evidence gate.
$1.000–$1.923 per GPU-hour · 2 regions.
Deployability evidence: 12/12 rows (100%). This is not a live-stock claim.
160 GPU-hours: $160.00–$307.60 observed compute range.
730 GPU-hours: $730.00–$1403.43 observed compute range.
$1.800–$1.800 per GPU-hour · 2 regions.
Deployability evidence: 6/6 rows (100%). This is not a live-stock claim.
160 GPU-hours: $288.00–$288.00 observed compute range.
730 GPU-hours: $1314.00–$1314.00 observed compute range.
$0.880–$1.558 per GPU-hour · 2 regions.
Deployability evidence: 18/18 rows (100%). This is not a live-stock claim.
160 GPU-hours: $140.80–$249.31 observed compute range.
730 GPU-hours: $642.40–$1137.48 observed compute range.
Start with Scaleway when provider choice is already constrained, then choose among the exact GPU models that currently pass the evidence gate.
$2.866–$2.866 per GPU-hour · 2 regions.
Deployability evidence: 4/4 rows (100%). This is not a live-stock claim.
160 GPU-hours: $458.64–$458.64 observed compute range.
730 GPU-hours: $2092.55–$2092.55 observed compute range.
$0.787–$0.787 per GPU-hour · 2 regions.
Deployability evidence: 8/8 rows (100%). This is not a live-stock claim.
160 GPU-hours: $126.00–$126.00 observed compute range.
730 GPU-hours: $574.88–$574.88 observed compute range.
Start with Vultr when provider choice is already constrained, then choose among the exact GPU models that currently pass the evidence gate.
$0.471–$0.471 per GPU-hour · 4 regions.
Deployability evidence: 5/5 rows (100%). This is not a live-stock claim.
160 GPU-hours: $75.36–$75.36 observed compute range.
730 GPU-hours: $343.83–$343.83 observed compute range.
Autonomous, not indiscriminate
Publish
A new URL becomes indexable only after its current official data passes family-specific breadth, freshness, source and uniqueness thresholds.
Update or noindex
Changed evidence triggers an update. A formerly qualified route that fails a cycle is preserved for users but immediately removed from search indexing.
Retire
After three consecutive failed cycles the route is retired. Canonical migrations are merged explicitly, preventing uncontrolled duplicates and crawl waste.