Blame for gpu_prices.csv

provider,gpu,variant,gpus_per_unit,price_usd_per_gpu_hr,price_type,term,region,billing_unit,price_as_listed,derivation,availability_note,lead_time_class,source_url,date_checked,fetch_method,confidence,notes
Latitude.sh,H100,80GB,1,1.68,on_demand,none,not itemized,"per hour, 1x H100 server",Starting at $1.68/hr,,Bare metal,unknown,https://www.latitude.sh/pricing,2026-11-01,manual,medium,'Starting at' price. Read by hand from the page on 2026-11-01. Region and plan dependent.
Voltage Park,H100,HGX,1,1.99,on_demand,none,not stated,per GPU-hour,starting at $1.99 per FAQ,,"Page: self-serve, up within 15 mins, no minimum terms",instant_self_serve,https://www.voltagepark.com/pricing,2026-11-09,html_static,low,Only a FAQ sentence gives a price ('starting at'). The pricing cards say Contact for pricing.
Hyperstack,H100,,1,2.50,on_demand,none,not stated,per GPU-hour (billed per minute),$2.50/hr,,,unknown,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,high,
Hyperstack,H100,NVL,1,2.60,on_demand,none,not stated,per GPU-hour (billed per minute),$2.60/hr,,,unknown,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,high,
Massed Compute,H100,80GB,1,2.73,on_demand,none,not stated,"per instance-hour, divided by GPUs",$2.73/hr for 1x,2.73 / 1,,instant_self_serve,https://www.massedcompute.com/pricing/,2026-11-09,html_static,medium,Page does not state region or availability. Prices are for the instance shown.
RunPod,H100,PCIe,1,2.89,on_demand,none,"30+ regions, not itemized",per GPU-hour (billed per second),$2.89/hr,,Pods page lists Community Cloud and Secure Cloud tabs. The static page text does not say which tier the shown price is.,instant_self_serve,https://www.runpod.io/pricing,2026-11-09,html_static,medium,Cloud tier of the displayed price is not stated in the page text. Reserved: contact enterprise sales.
Civo,H100,80GB,1,2.99,on_demand,none,not stated,"per instance-hour, divided by GPUs",$2.99 per hour for 1x,2.99 / 1,,instant_self_serve,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Massed Compute,H100,SXM5 80GB,1,3.14,on_demand,none,not stated,"per instance-hour, divided by GPUs",$3.14/hr for 1x,3.14 / 1,,instant_self_serve,https://www.massedcompute.com/pricing/,2026-11-09,html_static,medium,Page does not state region or availability. Prices are for the instance shown.
RunPod,H100,NVL,1,3.19,on_demand,none,"30+ regions, not itemized",per GPU-hour (billed per second),$3.19/hr,,Pods page lists Community Cloud and Secure Cloud tabs. The static page text does not say which tier the shown price is.,instant_self_serve,https://www.runpod.io/pricing,2026-11-09,html_static,medium,Cloud tier of the displayed price is not stated in the page text. Reserved: contact enterprise sales.
Hyperstack,H100,SXM,1,3.20,on_demand,none,not stated,per GPU-hour (billed per minute),$3.20/hr,,,unknown,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,high,
Thunder Compute,H100,SXM,1,3.20,on_demand,none,not stated,per GPU-hour (billed per minute),$3.20/hr,,"Page: persistent GPU machines, ready to SSH into in minutes",instant_self_serve,https://thundercompute.com/pricing,2026-11-09,html_static,high,
Lambda,H100,PCIe 80GB,1,3.79,on_demand,none,not stated,per GPU-hour,$3.79 per GPU-hr (1x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Crusoe,H100,80GB HGX,1,3.90,on_demand,none,not stated,per GPU-hour,$3.90/GPU-hr,,Page: contact sales to lock in guaranteed resources at lowest rates,unknown,https://www.crusoe.ai/cloud/pricing,2026-11-09,html_static,medium,Price column read by position (the page lists models and prices in separate blocks). Verify before relying on it.
RunPod,H100,SXM,1,3.99,on_demand,none,"30+ regions, not itemized",per GPU-hour (billed per second),$3.99/hr,,Pods page lists Community Cloud and Secure Cloud tabs. The static page text does not say which tier the shown price is.,instant_self_serve,https://www.runpod.io/pricing,2026-11-09,html_static,medium,Cloud tier of the displayed price is not stated in the page text. Reserved: contact enterprise sales.
Together AI,H100,HGX,1,3.99,on_demand,none,not stated,per GPU-hour,$3.99,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,GPU Clusters product. Page also shows a separate Dedicated Endpoints table with different prices.
DigitalOcean,H100,HGX,1,4.41,on_demand,none,not stated,"per GPU-hour (on-demand billed per second, 5 minute minimum)",$4.41/GPU/hour,,,instant_self_serve,https://www.digitalocean.com/pricing/gpu-droplets,2026-11-09,html_static,high,
Nebius,H100,HGX,1,4.50,on_demand,none,not stated,per GPU-hour,$4.50 per GPU-hr,,Reserved and committed: contact sales,unknown,https://www.nebius.com/prices,2026-11-09,html_static,high,
Lambda,H100,HGX 80GB,8,4.59,on_demand,none,not stated,per GPU-hour,$4.59 per GPU-hr (8x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Lambda,H100,SXM 80GB,4,4.69,on_demand,none,not stated,per GPU-hour,$4.69 per GPU-hr (4x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Lambda,H100,SXM 80GB,2,4.79,on_demand,none,not stated,per GPU-hour,$4.79 per GPU-hr (2x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Lambda,H100,SXM 80GB,1,4.89,on_demand,none,not stated,per GPU-hour,$4.89 per GPU-hr (1x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Verda,H100,SXM5 80GB,1,4.987,on_demand,none,not stated,per GPU-hour,$4.987/h,,"Reserved: from 1 month, custom quote (talk to sales)",instant_self_serve,https://verda.com/pricing,2026-11-09,html_static,high,
CoreWeave,H100,HGX,8,6.155,on_demand,none,North America,"per instance-hour, divided by GPUs",$49.24/hr per 8-GPU instance,49.24 / 8,Reserved and committed capacity: contact sales,unknown,https://www.coreweave.com/pricing,2026-11-09,html_static,high,
Amazon Web Services,H100,"SXM 80GB, 8x",8,6.88,on_demand,none,us-east-1,"per instance-hour, divided by GPUs","$55.04/hr for p5.48xlarge (Linux, shared tenancy)",55.04 / 8,Capacity Blocks for ML and quota limits apply; not shown in the price list,unknown,https://aws.amazon.com/ec2/pricing/on-demand/,2026-11-01,official_api,high,AWS Price List bulk file for AmazonEC2 us-east-1.
Amazon Web Services,H100,"80GB, 1x",1,6.88,on_demand,none,us-east-1,"per instance-hour, divided by GPUs","$6.88/hr for p5.4xlarge (Linux, shared tenancy)",6.88 / 1,Capacity Blocks for ML and quota limits apply; not shown in the price list,unknown,https://aws.amazon.com/ec2/pricing/on-demand/,2026-11-01,official_api,high,AWS Price List bulk file for AmazonEC2 us-east-1.
Microsoft Azure,H100,"NVL 94GB, 1x VM",1,6.98,on_demand,none,eastus,"per VM-hour, divided by GPUs",$6.98/hr for Standard_NC40ads_H100_v5 (Linux),6.98 / 1,Quota and capacity approval apply; not shown by the price API,unknown,https://learn.microsoft.com/en-us/rest/api/cost-management/retail-prices/azure-retail-prices,2026-11-09,official_api,high,"Retail list price, US dollars, Linux, pay as you go."
Oracle Cloud (OCI),H100,,1,10.00,on_demand,none,global list price,per GPU-hour,$10 GPU Per Hour (B98415),,,unknown,https://www.oracle.com/cloud/price-list/,2026-11-09,official_api,high,"Oracle Cloud Price List API, lastUpdated 2026-11-02T14:49:49.342Z. Part: Oracle Cloud Infrastructure - Compute - GPU - H100. GPU memory size is not stated in the price list. Capacity must be requested; not shown by the price list."
Microsoft Azure,H100,"SXM 80GB, 8x VM",8,12.29,on_demand,none,eastus,"per VM-hour, divided by GPUs",$98.32/hr for Standard_ND96isr_H100_v5 (Linux),98.32 / 8,Quota and capacity approval apply; not shown by the price API,unknown,https://learn.microsoft.com/en-us/rest/api/cost-management/retail-prices/azure-retail-prices,2026-11-09,official_api,high,"Retail list price, US dollars, Linux, pay as you go."
Modal,H100,SXM5,1,3.9492,serverless_per_second,none,not stated,"per second, shown x 3600",$0.001097/sec,0.001097 x 3600,,instant_self_serve,https://modal.com/pricing,2026-11-09,html_static,high,"Serverless billing, not a rented instance. Only pay while running."
Hyperstack,H100,,1,1.75,reserved,"reservation, term not stated on page",not stated,per GPU-hour,$1.75/hr (starting from),,,contact_sales,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,medium,Reserved price is 'starting from'. Term length is not shown.
Hyperstack,H100,NVL,1,1.82,reserved,"reservation, term not stated on page",not stated,per GPU-hour,$1.82/hr (starting from),,,contact_sales,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,medium,Reserved price is 'starting from'. Term length is not shown.
Civo,H100,80GB,1,2.49,reserved,36 months,not stated,"per instance-hour, divided by GPUs","$2.49 per hour for 1x, 36 months commitment",2.49 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,H100,80GB,1,2.59,reserved,24 months,not stated,"per instance-hour, divided by GPUs","$2.59 per hour for 1x, 24 months commitment",2.59 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,H100,80GB,1,2.69,reserved,12 months,not stated,"per instance-hour, divided by GPUs","$2.69 per hour for 1x, 12 months commitment",2.69 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Hyperstack,H100,SXM,1,2.72,reserved,"reservation, term not stated on page",not stated,per GPU-hour,$2.72/hr (starting from),,,contact_sales,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,medium,Reserved price is 'starting from'. Term length is not shown.
Civo,H100,80GB,1,2.79,reserved,6 months,not stated,"per instance-hour, divided by GPUs","$2.79 per hour for 1x, 6 months commitment",2.79 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
FlexiCloud (broker),H100,HGX SXM,8,2.95,reserved,"12 months or longer, floor rate on the longest term",by quote,"per GPU-hour, dedicated 8-GPU nodes, billed upfront for the term",from $2.95 per GPU-hour,,Page: available now; lead time per region is quoted in writing within about one working day; B300 allocation is the tightest,contact_sales,https://flexicloud.co/,2026-11-09,html_static,medium,"Broker quote floor, not an on-demand price. Minimum one 8-GPU node, one-year minimum. Actual rate depends on region, term and node count."
Together AI,H100,HGX,1,3.19,reserved,91 to 180 days,not stated,per GPU-hour,$3.19,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,
DigitalOcean,H100,HGX,1,3.26,reserved,12 months,not stated,"per GPU-hour (on-demand billed per second, 5 minute minimum)",$3.26/GPU/hour,,"Page: new pricing effective August 1, 2026. Reserved: contact sales.",contact_sales,https://www.digitalocean.com/pricing/gpu-droplets,2026-11-09,html_static,high,
Together AI,H100,HGX,1,3.45,reserved,31 to 90 days,not stated,per GPU-hour,$3.45,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,
Together AI,H100,HGX,1,3.69,reserved,3 to 30 days,not stated,per GPU-hour,$3.69,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,
Lambda,H100,,256,5.54,reserved,2 weeks to 1 year,,per GPU-hour,"$5.54 per GPU-hr at 256 GPUs, 2 weeks to 1 year",,1-Click Cluster: talk to sales,contact_sales,https://lambda.ai/pricing,2026-11-09,html_static,high,Cluster price includes the cluster network. Count is the minimum cluster size for that price.
Lambda,H100,,64,5.85,reserved,2 weeks to 1 year,,per GPU-hour,"$5.85 per GPU-hr at 64 GPUs, 2 weeks to 1 year",,1-Click Cluster: talk to sales,contact_sales,https://lambda.ai/pricing,2026-11-09,html_static,high,Cluster price includes the cluster network. Count is the minimum cluster size for that price.
Lambda,H100,,16,6.16,reserved,2 weeks to 1 year,,per GPU-hour,"$6.16 per GPU-hr at 16 GPUs, 2 weeks to 1 year",,1-Click Cluster: talk to sales,contact_sales,https://lambda.ai/pricing,2026-11-09,html_static,high,Cluster price includes the cluster network. Count is the minimum cluster size for that price.
Nebius,H100,HGX,1,0.79,preemptible,none,not stated,per GPU-hour,from $0.79 per GPU-hr (preemptible),,,unknown,https://www.nebius.com/prices,2026-11-09,html_static,medium,"Dynamic spot-like price, 'from' is the lowest seen at fetch time."
Microsoft Azure,H100,"NVL 94GB, 1x VM",1,1.4183,spot,none,eastus,"per VM-hour, divided by GPUs",$1.418336/hr for Standard_NC40ads_H100_v5 (Linux),1.418336 / 1,Quota and capacity approval apply; not shown by the price API,unknown,https://learn.microsoft.com/en-us/rest/api/cost-management/retail-prices/azure-retail-prices,2026-11-09,official_api,high,"Retail list price, US dollars, Linux, pay as you go."
Together AI,H100,HGX,1,1.99,preemptible,none,not stated,per GPU-hour,$1.99,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,
Hyperstack,H100,PCIe,1,2.00,spot,none,not stated,per GPU-hour,$2.00/hr (spot),,,unknown,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,medium,
Microsoft Azure,H100,"SXM 80GB, 8x VM",8,2.2712,spot,none,eastus,"per VM-hour, divided by GPUs",$18.169536/hr for Standard_ND96isr_H100_v5 (Linux),18.169536 / 8,Quota and capacity approval apply; not shown by the price API,unknown,https://learn.microsoft.com/en-us/rest/api/cost-management/retail-prices/azure-retail-prices,2026-11-09,official_api,high,"Retail list price, US dollars, Linux, pay as you go."
Verda,H100,SXM5 80GB,1,2.494,spot,none,not stated,per GPU-hour,$2.494/h (spot),,,unknown,https://verda.com/pricing,2026-11-09,html_static,high,
Fluidstack,H100,,,,contact_sales,,,,No public pricing page found (/pricing returns 404),,,contact_sales,https://www.fluidstack.io/,2026-11-01,manual,medium,"Quote only, as far as public pages show."
Together AI,H100,HGX,1,,contact_sales,181+ days,not stated,per GPU-hour,Contact us,,,contact_sales,https://www.together.ai/pricing,2026-11-09,html_static,high,
Voltage Park,H100,HGX,,,contact_sales,6+ months,not stated,,Contact for pricing,,"Page: Long-Term Reserve, 6+ month contracts, 32 to 8000+ H100 GPUs",contact_sales,https://www.voltagepark.com/pricing,2026-11-09,html_static,high,
Google Cloud,H100,,,,not_public,,,,Pricing page is rendered by script; static fetch returns no rates,,,,https://cloud.google.com/compute/gpus-pricing,2026-11-01,manual,high,"The Cloud Billing Catalog API needs an API key, so the refresh script does not call it. Not recorded. Check the page or the pricing calculator."
Hyperbolic,H100,,,,not_public,,,,"On-demand, reserved and private cloud offered. No rate in the home page text",,"FAQ: launch an on-demand GPU instance in minutes. Reserved: shorter-term commitment, usually under one year",instant_self_serve,https://hyperbolic.xyz/,2026-11-01,manual,medium,Rates are not in the fetched page text. Not recorded.
Paperspace (DigitalOcean),H100,,,,not_public,,,,Page shows $3.09/hour on the H100 card and a footnote saying $5.95/hour on-demand promo for H100,,,,https://www.paperspace.com/pricing,2026-11-01,manual,low,"The two figures conflict on the same page, so neither is recorded. Check the page."
SF Compute,H100,,,,not_public,,,,"Calculator on site, no static rate in page text",,,,https://sfcompute.com/,2026-11-01,manual,low,Public price feed reported stopped in late September 2026 (research/bigger-plays-entity-2026-11-01.md). Not verified again here.
Vast.ai,H100,,,,not_public,,,"per second, market price",Market-set. No static rate on the pricing page,,,instant_self_serve,https://vast.ai/pricing,2026-11-01,manual,high,Page says prices are set by supply and demand and shown live in the console. Marketplace hosts vary in reliability and verification. No number is recorded here.
Civo,H200,141GB,1,3.49,on_demand,none,not stated,"per instance-hour, divided by GPUs",$3.49 per hour for 1x,3.49 / 1,,instant_self_serve,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Massed Compute,H200,NVL 141GB,1,3.62,on_demand,none,not stated,"per instance-hour, divided by GPUs",$3.62/hr for 1x,3.62 / 1,,instant_self_serve,https://www.massedcompute.com/pricing/,2026-11-09,html_static,medium,Page does not state region or availability. Prices are for the instance shown.
Hyperstack,H200,SXM,1,3.99,on_demand,none,not stated,per GPU-hour (billed per minute),$3.99/hr,,,unknown,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,high,
Crusoe,H200,141GB HGX,1,4.29,on_demand,none,not stated,per GPU-hour,$4.29/GPU-hr,,Page: contact sales to lock in guaranteed resources at lowest rates,unknown,https://www.crusoe.ai/cloud/pricing,2026-11-09,html_static,medium,Price column read by position (the page lists models and prices in separate blocks). Verify before relying on it.
DigitalOcean,H200,HGX,1,4.47,on_demand,none,not stated,"per GPU-hour (on-demand billed per second, 5 minute minimum)",$4.47/GPU/hour,,,instant_self_serve,https://www.digitalocean.com/pricing/gpu-droplets,2026-11-09,html_static,high,
RunPod,H200,,1,5.29,on_demand,none,"30+ regions, not itemized",per GPU-hour (billed per second),$5.29/hr,,Pods page lists Community Cloud and Secure Cloud tabs. The static page text does not say which tier the shown price is.,instant_self_serve,https://www.runpod.io/pricing,2026-11-09,html_static,medium,Cloud tier of the displayed price is not stated in the page text. Reserved: contact enterprise sales.
Nebius,H200,HGX,1,5.40,on_demand,none,not stated,per GPU-hour,$5.40 per GPU-hr,,Reserved and committed: contact sales,unknown,https://www.nebius.com/prices,2026-11-09,html_static,high,
Together AI,H200,HGX,1,5.99,on_demand,none,not stated,per GPU-hour,$5.99,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,GPU Clusters product. Page also shows a separate Dedicated Endpoints table with different prices.
CoreWeave,H200,HGX,8,6.305,on_demand,none,North America,"per instance-hour, divided by GPUs",$50.44/hr per 8-GPU instance,50.44 / 8,Reserved and committed capacity: contact sales,unknown,https://www.coreweave.com/pricing,2026-11-09,html_static,high,
Verda,H200,SXM5 141GB,1,6.44,on_demand,none,not stated,per GPU-hour,$6.440/h,,"Reserved: from 1 month, custom quote (talk to sales)",instant_self_serve,https://verda.com/pricing,2026-11-09,html_static,high,
Amazon Web Services,H200,"SXM 141GB, 8x",8,7.912,on_demand,none,us-east-1,"per instance-hour, divided by GPUs","$63.296/hr for p5en.48xlarge (Linux, shared tenancy)",63.296 / 8,Capacity Blocks for ML and quota limits apply; not shown in the price list,unknown,https://aws.amazon.com/ec2/pricing/on-demand/,2026-11-01,official_api,high,AWS Price List bulk file for AmazonEC2 us-east-1.
Oracle Cloud (OCI),H200,,1,10.00,on_demand,none,global list price,per GPU-hour,$10 GPU Per Hour (B110519),,,unknown,https://www.oracle.com/cloud/price-list/,2026-11-09,official_api,high,"Oracle Cloud Price List API, lastUpdated 2026-11-02T14:49:49.342Z. Part: OCI - Compute - GPU - H200. GPU memory size is not stated in the price list. Capacity must be requested; not shown by the price list."
Modal,H200,SXM,1,4.5396,serverless_per_second,none,not stated,"per second, shown x 3600",$0.001261/sec,0.001261 x 3600,,instant_self_serve,https://modal.com/pricing,2026-11-09,html_static,high,"Serverless billing, not a rented instance. Only pay while running."
Hyperstack,H200,SXM,1,2.79,reserved,"reservation, term not stated on page",not stated,per GPU-hour,$2.79/hr (starting from),,,contact_sales,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,medium,Reserved price is 'starting from'. Term length is not shown.
Civo,H200,141GB,1,2.99,reserved,36 months,not stated,"per instance-hour, divided by GPUs","$2.99 per hour for 1x, 36 months commitment",2.99 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,H200,141GB,1,3.09,reserved,24 months,not stated,"per instance-hour, divided by GPUs","$3.09 per hour for 1x, 24 months commitment",3.09 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,H200,141GB,1,3.19,reserved,12 months,not stated,"per instance-hour, divided by GPUs","$3.19 per hour for 1x, 12 months commitment",3.19 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,H200,141GB,1,3.29,reserved,6 months,not stated,"per instance-hour, divided by GPUs","$3.29 per hour for 1x, 6 months commitment",3.29 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
FlexiCloud (broker),H200,HGX SXM,8,3.35,reserved,"12 months or longer, floor rate on the longest term",by quote,"per GPU-hour, dedicated 8-GPU nodes, billed upfront for the term",from $3.35 per GPU-hour,,Page: available now; lead time per region is quoted in writing within about one working day; B300 allocation is the tightest,contact_sales,https://flexicloud.co/,2026-11-09,html_static,medium,"Broker quote floor, not an on-demand price. Minimum one 8-GPU node, one-year minimum. Actual rate depends on region, term and node count."
DigitalOcean,H200,HGX,1,3.40,reserved,12 months,not stated,"per GPU-hour (on-demand billed per second, 5 minute minimum)",$3.40/GPU/hour,,"Page: new pricing effective August 1, 2026. Reserved: contact sales.",contact_sales,https://www.digitalocean.com/pricing/gpu-droplets,2026-11-09,html_static,high,
Together AI,H200,HGX,1,3.99,reserved,91 to 180 days,not stated,per GPU-hour,$3.99,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,
Together AI,H200,HGX,1,4.15,reserved,31 to 90 days,not stated,per GPU-hour,$4.15,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,
Together AI,H200,HGX,1,4.99,reserved,3 to 30 days,not stated,per GPU-hour,$4.99,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,
Nebius,H200,HGX,1,0.79,preemptible,none,not stated,per GPU-hour,from $0.79 per GPU-hr (preemptible),,,unknown,https://www.nebius.com/prices,2026-11-09,html_static,medium,"Dynamic spot-like price, 'from' is the lowest seen at fetch time."
Together AI,H200,HGX,1,2.99,preemptible,none,not stated,per GPU-hour,$2.99,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,
Verda,H200,SXM5 141GB,1,3.22,spot,none,not stated,per GPU-hour,$3.220/h (spot),,,unknown,https://verda.com/pricing,2026-11-09,html_static,high,
Together AI,H200,HGX,1,,contact_sales,181+ days,not stated,per GPU-hour,Contact us,,,contact_sales,https://www.together.ai/pricing,2026-11-09,html_static,high,
Hyperbolic,H200,,,,not_public,,,,No rate in the home page text,,,instant_self_serve,https://hyperbolic.xyz/,2026-11-01,manual,medium,See H100 row.
Massed Compute,B200,SXM6,8,5.4325,on_demand,none,not stated,"per instance-hour, divided by GPUs",$43.46/hr for 8x,43.46 / 8,,instant_self_serve,https://www.massedcompute.com/pricing/,2026-11-09,html_static,medium,Page does not state region or availability. Prices are for the instance shown.
Hyperstack,B200,,1,6.00,on_demand,none,not stated,per GPU-hour (billed per minute),$6.00/hr,,,unknown,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,high,
Lambda,B200,HGX 180GB,8,7.69,on_demand,none,not stated,per GPU-hour,$7.69 per GPU-hr (8x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Lambda,B200,SXM6 180GB,4,7.79,on_demand,none,not stated,per GPU-hour,$7.79 per GPU-hr (4x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Lambda,B200,SXM6 180GB,2,7.89,on_demand,none,not stated,per GPU-hour,$7.89 per GPU-hr (2x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Lambda,B200,SXM6 180GB,1,7.99,on_demand,none,not stated,per GPU-hour,$7.99 per GPU-hr (1x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
RunPod,B200,,1,7.99,on_demand,none,"30+ regions, not itemized",per GPU-hour (billed per second),$7.99/hr,,Pods page lists Community Cloud and Secure Cloud tabs. The static page text does not say which tier the shown price is.,instant_self_serve,https://www.runpod.io/pricing,2026-11-09,html_static,medium,Cloud tier of the displayed price is not stated in the page text. Reserved: contact enterprise sales.
Together AI,B200,HGX,1,8.19,on_demand,none,not stated,per GPU-hour,$8.19,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,GPU Clusters product. Page also shows a separate Dedicated Endpoints table with different prices.
Nebius,B200,HGX,1,8.50,on_demand,none,not stated,per GPU-hour,$8.50 per GPU-hr,,Reserved and committed: contact sales,unknown,https://www.nebius.com/prices,2026-11-09,html_static,high,
CoreWeave,B200,HGX,8,8.60,on_demand,none,North America,"per instance-hour, divided by GPUs",$68.80/hr per 8-GPU instance,68.8 / 8,Reserved and committed capacity: contact sales,unknown,https://www.coreweave.com/pricing,2026-11-09,html_static,high,
Verda,B200,SXM6 180GB,1,9.138,on_demand,none,not stated,per GPU-hour,$9.138/h,,"Reserved: from 1 month, custom quote (talk to sales)",instant_self_serve,https://verda.com/pricing,2026-11-09,html_static,high,
Oracle Cloud (OCI),B200,,1,14.00,on_demand,none,global list price,per GPU-hour,$14 GPU Per Hour (B110978),,,unknown,https://www.oracle.com/cloud/price-list/,2026-11-09,official_api,high,"Oracle Cloud Price List API, lastUpdated 2026-11-02T14:49:49.342Z. Part: OCI - Compute - GPU - B200. GPU memory size is not stated in the price list. Capacity must be requested; not shown by the price list."
Amazon Web Services,B200,SXM 8x,8,14.2416,on_demand,none,us-east-1,"per instance-hour, divided by GPUs","$113.9328/hr for p6-b200.48xlarge (Linux, shared tenancy)",113.9328 / 8,Capacity Blocks for ML and quota limits apply; not shown in the price list,unknown,https://aws.amazon.com/ec2/pricing/on-demand/,2026-11-01,official_api,high,AWS Price List bulk file for AmazonEC2 us-east-1.
Modal,B200,,1,6.2496,serverless_per_second,none,not stated,"per second, shown x 3600",$0.001736/sec,0.001736 x 3600,,instant_self_serve,https://modal.com/pricing,2026-11-09,html_static,high,"Serverless billing, not a rented instance. Only pay while running."
FlexiCloud (broker),B200,HGX SXM,8,3.75,reserved,"12 months or longer, floor rate on the longest term",by quote,"per GPU-hour, dedicated 8-GPU nodes, billed upfront for the term",from $3.75 per GPU-hour,,Page: available now; lead time per region is quoted in writing within about one working day; B300 allocation is the tightest,contact_sales,https://flexicloud.co/,2026-11-09,html_static,medium,"Broker quote floor, not an on-demand price. Minimum one 8-GPU node, one-year minimum. Actual rate depends on region, term and node count."
Civo,B200,180GB,8,3.79,reserved,36 months,not stated,"per instance-hour, divided by GPUs","$30.32 per hour for 8x, 36 months commitment",30.32 / 8,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,B200,180GB,8,3.99,reserved,24 months,not stated,"per instance-hour, divided by GPUs","$31.92 per hour for 8x, 24 months commitment",31.92 / 8,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,B200,180GB,8,4.49,reserved,12 months,not stated,"per instance-hour, divided by GPUs","$35.92 per hour for 8x, 12 months commitment",35.92 / 8,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Hyperstack,B200,,1,5.10,reserved,"reservation, term not stated on page",not stated,per GPU-hour,$5.10/hr (starting from),,,contact_sales,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,medium,Reserved price is 'starting from'. Term length is not shown.
Together AI,B200,HGX,1,6.79,reserved,91 to 180 days,not stated,per GPU-hour,$6.79,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,
Together AI,B200,HGX,1,7.79,reserved,31 to 90 days,not stated,per GPU-hour,$7.79,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,
Together AI,B200,HGX,1,7.99,reserved,3 to 30 days,not stated,per GPU-hour,$7.99,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,
Lambda,B200,HGX,256,8.87,reserved,2 weeks to 1 year,,per GPU-hour,"$8.87 per GPU-hr at 256+ GPUs, 2 weeks to 1 year",,1-Click Cluster: talk to sales,contact_sales,https://lambda.ai/pricing,2026-11-09,html_static,high,Cluster price includes the cluster network. Count is the minimum cluster size for that price.
Lambda,B200,HGX,64,9.36,reserved,2 weeks to 1 year,,per GPU-hour,"$9.36 per GPU-hr at 64 GPUs, 2 weeks to 1 year",,1-Click Cluster: talk to sales,contact_sales,https://lambda.ai/pricing,2026-11-09,html_static,high,Cluster price includes the cluster network. Count is the minimum cluster size for that price.
Lambda,B200,HGX,16,9.86,reserved,2 weeks to 1 year,,per GPU-hour,"$9.86 per GPU-hr at 16 GPUs, 2 weeks to 1 year",,1-Click Cluster: talk to sales,contact_sales,https://lambda.ai/pricing,2026-11-09,html_static,high,Cluster price includes the cluster network. Count is the minimum cluster size for that price.
Nebius,B200,HGX,1,0.99,preemptible,none,not stated,per GPU-hour,from $0.99 per GPU-hr (preemptible),,,unknown,https://www.nebius.com/prices,2026-11-09,html_static,medium,"Dynamic spot-like price, 'from' is the lowest seen at fetch time."
Together AI,B200,HGX,1,4.09,preemptible,none,not stated,per GPU-hour,$4.09,,,unknown,https://www.together.ai/pricing,2026-11-09,html_static,high,
Verda,B200,SXM6 180GB,1,4.569,spot,none,not stated,per GPU-hour,$4.569/h (spot),,,unknown,https://verda.com/pricing,2026-11-09,html_static,high,
Crusoe,B200,180GB HGX,,,contact_sales,,,,Contact sales,,,contact_sales,https://www.crusoe.ai/cloud/pricing,2026-11-09,html_static,high,
Together AI,B200,HGX,1,,contact_sales,181+ days,not stated,per GPU-hour,Contact us,,,contact_sales,https://www.together.ai/pricing,2026-11-09,html_static,high,
Google Cloud,B200,,,,not_public,,,,Pricing page is rendered by script; static fetch returns no rates,,,,https://cloud.google.com/compute/gpus-pricing,2026-11-01,manual,high,See H100 row.
Hyperbolic,B200,,,,not_public,,,,No rate in the home page text,,,instant_self_serve,https://hyperbolic.xyz/,2026-11-01,manual,medium,See H100 row.
Vast.ai,B200,,,,not_public,,,"per second, market price",Market-set. No static rate on the pricing page,,,instant_self_serve,https://vast.ai/pricing,2026-11-01,manual,high,Same as above.
Hyperstack,B300,,1,7.40,on_demand,none,not stated,per GPU-hour (billed per minute),$7.40/hr,,"Page header: B300 on-demand in August, reserved private clusters in Q4 (banner text, may be out of date)",unknown,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,high,
Civo,B300,,8,7.49,on_demand,none,not stated,"per instance-hour, divided by GPUs",$59.92 per hour for 8x,59.92 / 8,,instant_self_serve,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Crusoe,B300,288GB HGX,1,8.96,on_demand,none,not stated,per GPU-hour,$8.96/GPU-hr,,Page: contact sales to lock in guaranteed resources at lowest rates,unknown,https://www.crusoe.ai/cloud/pricing,2026-11-09,html_static,medium,Price column read by position (the page lists models and prices in separate blocks). Verify before relying on it.
RunPod,B300,,1,8.99,on_demand,none,"30+ regions, not itemized",per GPU-hour (billed per second),$8.99/hr,,Pods page lists Community Cloud and Secure Cloud tabs. The static page text does not say which tier the shown price is.,instant_self_serve,https://www.runpod.io/pricing,2026-11-09,html_static,medium,Cloud tier of the displayed price is not stated in the page text. Reserved: contact enterprise sales.
Nebius,B300,HGX,1,9.50,on_demand,none,not stated,per GPU-hour,$9.50 per GPU-hr,,Reserved and committed: contact sales,unknown,https://www.nebius.com/prices,2026-11-09,html_static,high,
Verda,B300,SXM6 268GB,1,11.81,on_demand,none,not stated,per GPU-hour,$11.81/h,,"Reserved: from 1 month, custom quote (talk to sales)",instant_self_serve,https://verda.com/pricing,2026-11-09,html_static,high,
Oracle Cloud (OCI),B300,,1,15.00,on_demand,none,global list price,per GPU-hour,$15 GPU Per Hour (B112237),,,unknown,https://www.oracle.com/cloud/price-list/,2026-11-09,official_api,high,"Oracle Cloud Price List API, lastUpdated 2026-11-02T14:49:49.342Z. Part: Oracle Cloud Infrastructure - Compute - GPU - B300. GPU memory size is not stated in the price list. Capacity must be requested; not shown by the price list."
Amazon Web Services,B300,SXM 8x,8,17.802,on_demand,none,us-east-1,"per instance-hour, divided by GPUs","$142.416/hr for p6-b300.48xlarge (Linux, shared tenancy)",142.416 / 8,Capacity Blocks for ML and quota limits apply; not shown in the price list,unknown,https://aws.amazon.com/ec2/pricing/on-demand/,2026-11-01,official_api,high,AWS Price List bulk file for AmazonEC2 us-east-1.
Modal,B300,,1,7.0992,serverless_per_second,none,not stated,"per second, shown x 3600",$0.001972/sec,0.001972 x 3600,,instant_self_serve,https://modal.com/pricing,2026-11-09,html_static,high,"Serverless billing, not a rented instance. Only pay while running."
FlexiCloud (broker),B300,HGX SXM,8,4.50,reserved,"12 months or longer, floor rate on the longest term",by quote,"per GPU-hour, dedicated 8-GPU nodes, billed upfront for the term",from $4.50 per GPU-hour,,Page: available now; lead time per region is quoted in writing within about one working day; B300 allocation is the tightest,contact_sales,https://flexicloud.co/,2026-11-09,html_static,medium,"Broker quote floor, not an on-demand price. Minimum one 8-GPU node, one-year minimum. Actual rate depends on region, term and node count."
Civo,B300,,8,5.49,reserved,36 months,not stated,"per instance-hour, divided by GPUs","$43.92 per hour for 8x, 36 months commitment",43.92 / 8,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,B300,,8,5.99,reserved,24 months,not stated,"per instance-hour, divided by GPUs","$47.92 per hour for 8x, 24 months commitment",47.92 / 8,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,B300,,8,6.49,reserved,12 months,not stated,"per instance-hour, divided by GPUs","$51.92 per hour for 8x, 12 months commitment",51.92 / 8,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
DigitalOcean,B300,HGX,1,7.94,reserved,12 months,not stated,"per GPU-hour (on-demand billed per second, 5 minute minimum)",$7.94/GPU/hour,,"Page: new pricing effective August 1, 2026. Reserved: contact sales.",contact_sales,https://www.digitalocean.com/pricing/gpu-droplets,2026-11-09,html_static,high,
Nebius,B300,HGX,1,0.99,preemptible,none,not stated,per GPU-hour,from $0.99 per GPU-hr (preemptible),,,unknown,https://www.nebius.com/prices,2026-11-09,html_static,medium,"Dynamic spot-like price, 'from' is the lowest seen at fetch time."
Verda,B300,SXM6 268GB,1,5.905,spot,none,not stated,per GPU-hour,$5.905/h (spot),,,unknown,https://verda.com/pricing,2026-11-09,html_static,high,
DigitalOcean,B300,HGX,1,6.17,spot,none,not stated,"per GPU-hour (on-demand billed per second, 5 minute minimum)",$6.17/GPU/hour,,,instant_self_serve,https://www.digitalocean.com/pricing/gpu-droplets,2026-11-09,html_static,high,
CoreWeave,B300,HGX,,,contact_sales,,North America,,Contact sales,,,contact_sales,https://www.coreweave.com/pricing,2026-11-09,html_static,high,
CoreWeave,GB200,NVL72,4,10.50,on_demand,none,North America,"per instance-hour, divided by GPUs",$42.00/hr per 4-GPU instance,42.0 / 4,Reserved and committed capacity: contact sales,unknown,https://www.coreweave.com/pricing,2026-11-09,html_static,high,
Oracle Cloud (OCI),GB200,,1,16.00,on_demand,none,global list price,per GPU-hour,$16 GPU Per Hour (B110979),,,unknown,https://www.oracle.com/cloud/price-list/,2026-11-09,official_api,high,"Oracle Cloud Price List API, lastUpdated 2026-11-02T14:49:49.342Z. Part: OCI - Compute - GPU - GB200. GPU memory size is not stated in the price list. Capacity must be requested; not shown by the price list."
Microsoft Azure,GB200,"NVL72, 4x VM",4,27.04,on_demand,none,eastus,"per VM-hour, divided by GPUs",$108.16/hr for Standard_ND128isr_NDR_GB200_v6 (Linux),108.16 / 4,Quota and capacity approval apply; not shown by the price API,unknown,https://learn.microsoft.com/en-us/rest/api/cost-management/retail-prices/azure-retail-prices,2026-11-09,official_api,medium,"GPU count per VM is from Azure VM size docs (ND GB200-v6 has 4 GPUs), not from the price API. Retail list price, US dollars, Linux, pay as you go."
Crusoe,GB200,186GB NVL72,,,contact_sales,,,,Contact sales,,,contact_sales,https://www.crusoe.ai/cloud/pricing,2026-11-09,html_static,high,
Nebius,GB200,NVL72,,,contact_sales,,,,Contact us,,,contact_sales,https://www.nebius.com/prices,2026-11-09,html_static,high,
Verda,GB300,SXM6 288GB,1,13.58,on_demand,none,not stated,per GPU-hour,$13.58/h,,"Reserved: from 1 month, custom quote (talk to sales)",instant_self_serve,https://verda.com/pricing,2026-11-09,html_static,high,
Oracle Cloud (OCI),GB300,,1,18.00,on_demand,none,global list price,per GPU-hour,$18 GPU Per Hour (B112140),,,unknown,https://www.oracle.com/cloud/price-list/,2026-11-09,official_api,high,"Oracle Cloud Price List API, lastUpdated 2026-11-02T14:49:49.342Z. Part: OCI - Compute - GPU - GB300. GPU memory size is not stated in the price list. Capacity must be requested; not shown by the price list."
Verda,GB300,SXM6 288GB,1,6.791,spot,none,not stated,per GPU-hour,$6.791/h (spot),,,unknown,https://verda.com/pricing,2026-11-09,html_static,high,
CoreWeave,GB300,NVL72,,,contact_sales,,North America,,Contact sales,,,contact_sales,https://www.coreweave.com/pricing,2026-11-09,html_static,high,
Crusoe,GB300,288GB NVL72,,,contact_sales,,,,Contact sales,,,contact_sales,https://www.crusoe.ai/cloud/pricing,2026-11-09,html_static,high,
Nebius,GB300,NVL72,,,contact_sales,,,,Contact us,,,contact_sales,https://www.nebius.com/prices,2026-11-09,html_static,high,
Civo,A100,40GB,1,1.09,on_demand,none,not stated,"per instance-hour, divided by GPUs",$1.09 per hour for 1x,1.09 / 1,,instant_self_serve,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Thunder Compute,A100,80GB,1,1.09,on_demand,none,not stated,per GPU-hour (billed per minute),$1.09/hr,,"Page: persistent GPU machines, ready to SSH into in minutes",instant_self_serve,https://thundercompute.com/pricing,2026-11-09,html_static,high,
Hyperstack,A100,,1,1.35,on_demand,none,not stated,per GPU-hour (billed per minute),$1.35/hr,,,unknown,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,high,
Massed Compute,A100,80GB,1,1.35,on_demand,none,not stated,"per instance-hour, divided by GPUs",$1.35/hr for 1x,1.35 / 1,,instant_self_serve,https://www.massedcompute.com/pricing/,2026-11-09,html_static,medium,Page does not state region or availability. Prices are for the instance shown.
Massed Compute,A100,SXM4 80GB,1,1.38,on_demand,none,not stated,"per instance-hour, divided by GPUs",$1.38/hr for 1x,1.38 / 1,,instant_self_serve,https://www.massedcompute.com/pricing/,2026-11-09,html_static,medium,Page does not state region or availability. Prices are for the instance shown.
Hyperstack,A100,NVL,1,1.40,on_demand,none,not stated,per GPU-hour (billed per minute),$1.40/hr,,,unknown,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,high,
Hyperstack,A100,SXM,1,1.60,on_demand,none,not stated,per GPU-hour (billed per minute),$1.60/hr,,,unknown,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,high,
Verda,A100,SXM4 40GB,1,1.657,on_demand,none,not stated,per GPU-hour,$1.657/h,,"Reserved: from 1 month, custom quote (talk to sales)",instant_self_serve,https://verda.com/pricing,2026-11-09,html_static,high,
RunPod,A100,PCIe,1,1.79,on_demand,none,"30+ regions, not itemized",per GPU-hour (billed per second),$1.79/hr,,Pods page lists Community Cloud and Secure Cloud tabs. The static page text does not say which tier the shown price is.,instant_self_serve,https://www.runpod.io/pricing,2026-11-09,html_static,medium,Cloud tier of the displayed price is not stated in the page text. Reserved: contact enterprise sales.
RunPod,A100,SXM,1,1.79,on_demand,none,"30+ regions, not itemized",per GPU-hour (billed per second),$1.79/hr,,Pods page lists Community Cloud and Secure Cloud tabs. The static page text does not say which tier the shown price is.,instant_self_serve,https://www.runpod.io/pricing,2026-11-09,html_static,medium,Cloud tier of the displayed price is not stated in the page text. Reserved: contact enterprise sales.
Lambda,A100,HGX 40GB,8,1.99,on_demand,none,not stated,per GPU-hour,$1.99 per GPU-hr (8x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Lambda,A100,PCIe 40GB,4,1.99,on_demand,none,not stated,per GPU-hour,$1.99 per GPU-hr (4x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Lambda,A100,PCIe 40GB,2,1.99,on_demand,none,not stated,per GPU-hour,$1.99 per GPU-hr (2x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Lambda,A100,SXM 40GB,1,1.99,on_demand,none,not stated,per GPU-hour,$1.99 per GPU-hr (1x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Lambda,A100,PCIe 40GB,1,1.99,on_demand,none,not stated,per GPU-hour,$1.99 per GPU-hr (1x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Crusoe,A100,80GB PCIe,1,2.00,on_demand,none,not stated,per GPU-hour,$2.00/GPU-hr,,Page: contact sales to lock in guaranteed resources at lowest rates,unknown,https://www.crusoe.ai/cloud/pricing,2026-11-09,html_static,medium,Price column read by position (the page lists models and prices in separate blocks). Verify before relying on it.
Crusoe,A100,80GB SXM,1,2.30,on_demand,none,not stated,per GPU-hour,$2.30/GPU-hr,,Page: contact sales to lock in guaranteed resources at lowest rates,unknown,https://www.crusoe.ai/cloud/pricing,2026-11-09,html_static,medium,Price column read by position (the page lists models and prices in separate blocks). Verify before relying on it.
Verda,A100,SXM4 80GB,1,2.316,on_demand,none,not stated,per GPU-hour,$2.316/h,,"Reserved: from 1 month, custom quote (talk to sales)",instant_self_serve,https://verda.com/pricing,2026-11-09,html_static,high,
CoreWeave,A100,,8,2.70,on_demand,none,North America,"per instance-hour, divided by GPUs",$21.60/hr per 8-GPU instance,21.6 / 8,Reserved and committed capacity: contact sales,unknown,https://www.coreweave.com/pricing,2026-11-09,html_static,high,
Amazon Web Services,A100,"SXM 40GB, 8x",8,2.7447,on_demand,none,us-east-1,"per instance-hour, divided by GPUs","$21.957642/hr for p4d.24xlarge (Linux, shared tenancy)",21.957642 / 8,Capacity Blocks for ML and quota limits apply; not shown in the price list,unknown,https://aws.amazon.com/ec2/pricing/on-demand/,2026-11-01,official_api,high,AWS Price List bulk file for AmazonEC2 us-east-1.
Lambda,A100,HGX 80GB,8,3.19,on_demand,none,not stated,per GPU-hour,$3.19 per GPU-hr (8x instance),,"Instances: page says deploy in minutes, self-serve, first-come access. Reserved and clusters: contact sales.",instant_self_serve,https://lambda.ai/pricing,2026-11-09,html_static,high,
Microsoft Azure,A100,"SXM 40GB, 8x VM",8,3.3996,on_demand,none,eastus,"per VM-hour, divided by GPUs",$27.197/hr for Standard_ND96asr_v4 (Linux),27.197 / 8,Quota and capacity approval apply; not shown by the price API,unknown,https://learn.microsoft.com/en-us/rest/api/cost-management/retail-prices/azure-retail-prices,2026-11-09,official_api,high,"Retail list price, US dollars, Linux, pay as you go."
Amazon Web Services,A100,"SXM 80GB, 8x",8,3.4309,on_demand,none,us-east-1,"per instance-hour, divided by GPUs","$27.44705/hr for p4de.24xlarge (Linux, shared tenancy)",27.44705 / 8,Capacity Blocks for ML and quota limits apply; not shown in the price list,unknown,https://aws.amazon.com/ec2/pricing/on-demand/,2026-11-01,official_api,high,AWS Price List bulk file for AmazonEC2 us-east-1.
Oracle Cloud (OCI),A100,,1,4.00,on_demand,none,global list price,per GPU-hour,$4 GPU Per Hour (B95907),,,unknown,https://www.oracle.com/cloud/price-list/,2026-11-09,official_api,high,"Oracle Cloud Price List API, lastUpdated 2026-11-02T14:49:49.342Z. Part: Compute - GPU - A100 - v2. GPU memory size is not stated in the price list. Capacity must be requested; not shown by the price list."
Microsoft Azure,A100,"SXM 80GB, 8x VM",8,4.0963,on_demand,none,eastus,"per VM-hour, divided by GPUs",$32.77/hr for Standard_ND96amsr_A100_v4 (Linux),32.77 / 8,Quota and capacity approval apply; not shown by the price API,unknown,https://learn.microsoft.com/en-us/rest/api/cost-management/retail-prices/azure-retail-prices,2026-11-09,official_api,high,"Retail list price, US dollars, Linux, pay as you go."
Modal,A100,40GB,1,2.0988,serverless_per_second,none,not stated,"per second, shown x 3600",$0.000583/sec,0.000583 x 3600,,instant_self_serve,https://modal.com/pricing,2026-11-09,html_static,high,"Serverless billing, not a rented instance. Only pay while running."
Modal,A100,80GB,1,2.4984,serverless_per_second,none,not stated,"per second, shown x 3600",$0.000694/sec,0.000694 x 3600,,instant_self_serve,https://modal.com/pricing,2026-11-09,html_static,high,"Serverless billing, not a rented instance. Only pay while running."
Civo,A100,40GB,1,0.69,reserved,36 months,not stated,"per instance-hour, divided by GPUs","$0.69 per hour for 1x, 36 months commitment",0.69 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,A100,40GB,1,0.79,reserved,24 months,not stated,"per instance-hour, divided by GPUs","$0.79 per hour for 1x, 24 months commitment",0.79 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,A100,40GB,1,0.89,reserved,12 months,not stated,"per instance-hour, divided by GPUs","$0.89 per hour for 1x, 12 months commitment",0.89 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Hyperstack,A100,,1,0.95,reserved,"reservation, term not stated on page",not stated,per GPU-hour,$0.95/hr (starting from),,,contact_sales,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,medium,Reserved price is 'starting from'. Term length is not shown.
Hyperstack,A100,NVL,1,0.98,reserved,"reservation, term not stated on page",not stated,per GPU-hour,$0.98/hr (starting from),,,contact_sales,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,medium,Reserved price is 'starting from'. Term length is not shown.
Civo,A100,40GB,1,0.99,reserved,6 months,not stated,"per instance-hour, divided by GPUs","$0.99 per hour for 1x, 6 months commitment",0.99 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Hyperstack,A100,SXM,1,1.36,reserved,"reservation, term not stated on page",not stated,per GPU-hour,$1.36/hr (starting from),,,contact_sales,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,medium,Reserved price is 'starting from'. Term length is not shown.
Microsoft Azure,A100,"SXM 40GB, 8x VM",8,0.7381,spot,none,eastus,"per VM-hour, divided by GPUs",$5.904469/hr for Standard_ND96asr_v4 (Linux),5.904469 / 8,Quota and capacity approval apply; not shown by the price API,unknown,https://learn.microsoft.com/en-us/rest/api/cost-management/retail-prices/azure-retail-prices,2026-11-09,official_api,high,"Retail list price, US dollars, Linux, pay as you go."
Verda,A100,SXM4 40GB,1,0.8285,spot,none,not stated,per GPU-hour,$0.8285/h (spot),,,unknown,https://verda.com/pricing,2026-11-09,html_static,high,
Microsoft Azure,A100,"SXM 80GB, 8x VM",8,1.0552,spot,none,eastus,"per VM-hour, divided by GPUs",$8.441552/hr for Standard_ND96amsr_A100_v4 (Linux),8.441552 / 8,Quota and capacity approval apply; not shown by the price API,unknown,https://learn.microsoft.com/en-us/rest/api/cost-management/retail-prices/azure-retail-prices,2026-11-09,official_api,high,"Retail list price, US dollars, Linux, pay as you go."
Hyperstack,A100,,1,1.08,spot,none,not stated,per GPU-hour,$1.08/hr (spot),,,unknown,https://www.hyperstack.cloud/gpu-pricing,2026-11-09,html_static,medium,
Verda,A100,SXM4 80GB,1,1.158,spot,none,not stated,per GPU-hour,$1.158/h (spot),,,unknown,https://verda.com/pricing,2026-11-09,html_static,high,
Massed Compute,L40S,48GB,1,0.97,on_demand,none,not stated,"per instance-hour, divided by GPUs",$0.97/hr for 1x,0.97 / 1,,instant_self_serve,https://www.massedcompute.com/pricing/,2026-11-09,html_static,medium,Page does not state region or availability. Prices are for the instance shown.
RunPod,L40S,,1,1.09,on_demand,none,"30+ regions, not itemized",per GPU-hour (billed per second),$1.09/hr,,Pods page lists Community Cloud and Secure Cloud tabs. The static page text does not say which tier the shown price is.,instant_self_serve,https://www.runpod.io/pricing,2026-11-09,html_static,medium,Cloud tier of the displayed price is not stated in the page text. Reserved: contact enterprise sales.
Civo,L40S,48GB,1,1.29,on_demand,none,not stated,"per instance-hour, divided by GPUs",$1.29 per hour for 1x,1.29 / 1,,instant_self_serve,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Crusoe,L40S,48GB,1,1.50,on_demand,none,not stated,per GPU-hour,$1.50/GPU-hr,,Page: contact sales to lock in guaranteed resources at lowest rates,unknown,https://www.crusoe.ai/cloud/pricing,2026-11-09,html_static,medium,Price column read by position (the page lists models and prices in separate blocks). Verify before relying on it.
Nebius,L40S,Intel CPU,1,1.55,on_demand,none,not stated,per GPU-hour,from $1.55 per GPU-hr,,Reserved and committed: contact sales,unknown,https://www.nebius.com/prices,2026-11-09,html_static,high,'from' means the lowest configuration shown.
DigitalOcean,L40S,,1,1.57,on_demand,none,not stated,"per GPU-hour (on-demand billed per second, 5 minute minimum)",$1.57/GPU/hour,,,instant_self_serve,https://www.digitalocean.com/pricing/gpu-droplets,2026-11-09,html_static,high,
Nebius,L40S,AMD CPU,1,1.82,on_demand,none,not stated,per GPU-hour,from $1.82 per GPU-hr,,Reserved and committed: contact sales,unknown,https://www.nebius.com/prices,2026-11-09,html_static,high,'from' means the lowest configuration shown.
Amazon Web Services,L40S,"48GB, 1x",1,1.861,on_demand,none,us-east-1,"per instance-hour, divided by GPUs","$1.861/hr for g6e.xlarge (Linux, shared tenancy)",1.861 / 1,Capacity Blocks for ML and quota limits apply; not shown in the price list,unknown,https://aws.amazon.com/ec2/pricing/on-demand/,2026-11-01,official_api,high,AWS Price List bulk file for AmazonEC2 us-east-1.
Verda,L40S,48GB,1,1.958,on_demand,none,not stated,per GPU-hour,$1.958/h,,"Reserved: from 1 month, custom quote (talk to sales)",instant_self_serve,https://verda.com/pricing,2026-11-09,html_static,high,
CoreWeave,L40S,,8,2.25,on_demand,none,North America,"per instance-hour, divided by GPUs",$18.00/hr per 8-GPU instance,18.0 / 8,Reserved and committed capacity: contact sales,unknown,https://www.coreweave.com/pricing,2026-11-09,html_static,high,
Oracle Cloud (OCI),L40S,,1,3.50,on_demand,none,global list price,per GPU-hour,$3.5 GPU Per Hour (B109479),,,unknown,https://www.oracle.com/cloud/price-list/,2026-11-09,official_api,high,"Oracle Cloud Price List API, lastUpdated 2026-11-02T14:49:49.342Z. Part: Compute - GPU - L40S. GPU memory size is not stated in the price list. Capacity must be requested; not shown by the price list."
Amazon Web Services,L40S,"48GB, 8x",8,3.7664,on_demand,none,us-east-1,"per instance-hour, divided by GPUs","$30.13118/hr for g6e.48xlarge (Linux, shared tenancy)",30.13118 / 8,Capacity Blocks for ML and quota limits apply; not shown in the price list,unknown,https://aws.amazon.com/ec2/pricing/on-demand/,2026-11-01,official_api,high,AWS Price List bulk file for AmazonEC2 us-east-1.
Modal,L40S,,1,1.9512,serverless_per_second,none,not stated,"per second, shown x 3600",$0.000542/sec,0.000542 x 3600,,instant_self_serve,https://modal.com/pricing,2026-11-09,html_static,high,"Serverless billing, not a rented instance. Only pay while running."
Civo,L40S,48GB,1,0.89,reserved,36 months,not stated,"per instance-hour, divided by GPUs","$0.89 per hour for 1x, 36 months commitment",0.89 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,L40S,48GB,1,0.99,reserved,24 months,not stated,"per instance-hour, divided by GPUs","$0.99 per hour for 1x, 24 months commitment",0.99 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,L40S,48GB,1,1.09,reserved,12 months,not stated,"per instance-hour, divided by GPUs","$1.09 per hour for 1x, 12 months commitment",1.09 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Civo,L40S,48GB,1,1.19,reserved,6 months,not stated,"per instance-hour, divided by GPUs","$1.19 per hour for 1x, 6 months commitment",1.19 / 1,,unknown,https://www.civo.com/pricing,2026-11-09,html_static,high,Small and Extra Large sizes only are read. N/A cells are skipped.
Nebius,L40S,Intel CPU,1,0.74,preemptible,none,not stated,per GPU-hour,from $0.74** per GPU-hr (preemptible),,,unknown,https://www.nebius.com/prices,2026-11-09,html_static,medium,"Dynamic spot-like price, 'from' is the lowest seen at fetch time."
Nebius,L40S,AMD CPU,1,0.90,preemptible,none,not stated,per GPU-hour,from $0.90** per GPU-hr (preemptible),,,unknown,https://www.nebius.com/prices,2026-11-09,html_static,medium,"Dynamic spot-like price, 'from' is the lowest seen at fetch time."
Verda,L40S,48GB,1,0.979,spot,none,not stated,per GPU-hour,$0.9790/h (spot),,,unknown,https://verda.com/pricing,2026-11-09,html_static,high,
Crusoe,MI300X,192GB,1,3.45,on_demand,none,not stated,per GPU-hour,$3.45/GPU-hr,,Page: contact sales to lock in guaranteed resources at lowest rates,unknown,https://www.crusoe.ai/cloud/pricing,2026-11-09,html_static,medium,Price column read by position (the page lists models and prices in separate blocks). Verify before relying on it.
Oracle Cloud (OCI),MI300X,,1,6.00,on_demand,none,global list price,per GPU-hour,$6 GPU Per Hour (B109485),,,unknown,https://www.oracle.com/cloud/price-list/,2026-11-09,official_api,high,"Oracle Cloud Price List API, lastUpdated 2026-11-02T14:49:49.342Z. Part: OCI - Compute - GPU - MI300X. GPU memory size is not stated in the price list. Capacity must be requested; not shown by the price list."
Oracle Cloud (OCI),MI355X,,1,8.60,on_demand,none,global list price,per GPU-hour,$8.6 GPU Per Hour (B111758),,,unknown,https://www.oracle.com/cloud/price-list/,2026-11-09,official_api,high,"Oracle Cloud Price List API, lastUpdated 2026-11-02T14:49:49.342Z. Part: OCI- Compute - GPU - MI355X. GPU memory size is not stated in the price list. Capacity must be requested; not shown by the price list."
Crusoe,MI355X,288GB,,,contact_sales,,,,Contact sales,,,contact_sales,https://www.crusoe.ai/cloud/pricing,2026-11-09,html_static,high,