Compare GPU prices across providers. Every price links to its official source and shows when it was last verified.
55 matching offerings · 86 billing options · filtered
| GPU | Provider | Type | Plan | Price | Location |
|---|---|---|---|---|---|
| NVIDIA B200 180 GB VRAM/GPU · fixed | Baseten.co B200 | Serverless Dedicated Deployments | B200 standard scale to zero · dedicated 1 minute increments · bills deploying, scaling up, scaling down, making predictions |
Component price — not a total | unknown |
| | |||||
| H100 80 GB VRAM/GPU · fixed | Baseten.co H100 | Serverless Dedicated Deployments | H100 standard scale to zero · dedicated 1 minute increments · bills deploying, scaling up, scaling down, making predictions |
Component price — not a total | unknown |
| | |||||
| A100 80 GB VRAM/GPU · fixed | Baseten.co A100 | Serverless Dedicated Deployments | A100 standard scale to zero · dedicated 1 minute increments · bills deploying, scaling up, scaling down, making predictions |
Component price — not a total | unknown |
| | |||||
| 1× NVIDIA B200 180 GB VRAM/GPU · fixed | Beam.cloud B200 | Serverless Serverless | Standard scale to zero 1 ms increments · bills code running |
Component price — not a total | unknown |
| The Serverless committed-spend option is presented as “With committed spend Book a call →” without a published amount; it was omitted.The Serverless section includes a “With committed spend” / “Book a call” option without a published amount; it was omitted. | |||||
| 1× A100 80 GB VRAM/GPU · fixed | Beam.cloud A100 80GB | Serverless Serverless | Standard scale to zero 1 ms increments · bills code running |
Component price — not a total | unknown |
| The Serverless committed-spend option is presented as “With committed spend Book a call →” without a published amount; it was omitted.The Serverless section includes a “With committed spend” / “Book a call” option without a published amount; it was omitted. | |||||
| 1× H100 80 GB VRAM/GPU · fixed | Beam.cloud H100 | Serverless Serverless | Standard serverless scale to zero 1 ms increments · bills code running |
Component price — not a total | unknown |
| The Serverless section includes a “With committed spend” / “Book a call” option without a published amount; it was omitted. | |||||
| 1× NVIDIA RTX PRO 6000 Blackwell 96 GB VRAM/GPU · fixed | Beam.cloud RTX PRO 6000 | Serverless Serverless | Standard scale to zero 1 ms increments · bills code running |
Component price — not a total | unknown |
| The Serverless committed-spend option is presented as “With committed spend Book a call →” without a published amount; it was omitted.The Serverless section includes a “With committed spend” / “Book a call” option without a published amount; it was omitted. | |||||
| 1× NVIDIA H200 141 GB VRAM/GPU · fixed | Beam.cloud H200 | Serverless Serverless | Standard scale to zero 1 ms increments · bills code running |
Component price — not a total | unknown |
| The Serverless committed-spend option is presented as “With committed spend Book a call →” without a published amount; it was omitted.The Serverless section includes a “With committed spend” / “Book a call” option without a published amount; it was omitted. | |||||
| 1× A100 80 GB VRAM/GPU · fixed | Cerebrium A100 (80GB) | Serverless Serverless | Standard scale to zero 1 second increments · bills active compute time |
Component price — not a total | unknown |
| Enterprise pricing is listed as Custom and was omitted. 1 geography evidence item
| |||||
| 1× A100 80 GB VRAM/GPU · fixed | Cerebrium A100 (80GB) | Serverless Serverless | Standard scale-to-zero scale to zero 1 second increments |
Component price — not a total | unknown |
| Enterprise pricing is listed as Custom and was omitted. 1 geography evidence item
| |||||
| 1× NVIDIA B200 180 GB VRAM/GPU · fixed | Fal.ai B200 | Serverless Serverless | Always on always on 1 second increments · bills setup, idle, running, draining, terminating |
≈ EUR 5.504/hour | unknown |
| Scale to zero scale to zero 1 second increments · bills setup, idle, running, draining, terminating |
≈ EUR 5.504/hour | ||||
| The source lists discounted “As low as” prices, but those custom deployment prices are omitted because the terms require sales contact. | |||||
| 1× NVIDIA B300 288 GB VRAM/GPU · fixed | Fal.ai B300 | Serverless Serverless | Always on always on 1 second increments · bills setup, idle, running, draining, terminating |
≈ EUR 7.486/hour | unknown |
| Scale to zero scale to zero 1 second increments · bills setup, idle, running, draining, terminating |
≈ EUR 7.486/hour | ||||
| The source lists discounted “As low as” prices, but those custom deployment prices are omitted because the terms require sales contact. | |||||
| 1× H100 80 GB VRAM/GPU · fixed | Fal.ai H100 | Serverless Serverless | Always on always on 1 second increments · bills setup, idle, running, draining, terminating |
≈ EUR 3.963/hour | unknown |
| Scale to zero scale to zero 1 second increments · bills setup, idle, running, draining, terminating |
≈ EUR 3.963/hour | ||||
| The source lists discounted “As low as” prices, but those custom deployment prices are omitted because the terms require sales contact. | |||||
| 1× NVIDIA RTX PRO 6000 Blackwell 96 GB VRAM/GPU · fixed | Fal.ai RTX PRO 6000 | Serverless Serverless | Always on always on 1 second increments · bills setup, idle, running, draining, terminating |
≈ EUR 2.633/hour | unknown |
| Scale to zero scale to zero 1 second increments · bills setup, idle, running, draining, terminating |
≈ EUR 2.633/hour | ||||
| The source lists discounted “As low as” prices, but those custom deployment prices are omitted because the terms require sales contact. | |||||
| 1× NVIDIA H200 141 GB VRAM/GPU · fixed | Fal.ai H200 | Serverless Serverless | Always on always on 1 second increments · bills setup, idle, running, draining, terminating |
≈ EUR 3.963/hour | unknown |
| Scale to zero scale to zero 1 second increments · bills setup, idle, running, draining, terminating |
≈ EUR 3.963/hour | ||||
| The source lists discounted “As low as” prices, but those custom deployment prices are omitted because the terms require sales contact. | |||||
| 1× NVIDIA RTX PRO 6000 Blackwell 96 GB VRAM/GPU · fixed | Google Cloud NVIDIA RTX Pro 6000 | Serverless Cloud Run GPU Services | NVIDIA RTX Pro 6000 No zonal redundancy scale to zero 100 ms increments · 1 minute minimum · bills entire instance lifecycle |
Component price — not a total | unknown |
| NVIDIA RTX Pro 6000 Zonal redundancy scale to zero 100 ms increments · 1 minute minimum · bills entire instance lifecycle |
Component price — not a total | ||||
1 geography evidence item
| |||||
| 1× NVIDIA RTX PRO 6000 Blackwell 96 GB VRAM/GPU · fixed | Google Cloud NVIDIA RTX Pro 6000 | Serverless Cloud Run GPU Services | NVIDIA RTX Pro 6000 No zonal redundancy scale to zero 100 ms increments · 1 minute minimum · bills entire instance lifecycle |
Component price — not a total | EU compute EU-constrained Netherlands |
| NVIDIA RTX Pro 6000 Zonal redundancy scale to zero 100 ms increments · 1 minute minimum · bills entire instance lifecycle |
Component price — not a total | ||||
3 geography evidence items
| |||||
| 1× A100 80 GB VRAM/GPU · fixed | Inferless Nvidia A100 Dedicated | Serverless Serverless | Standard scale to zero · dedicated 1 second increments · bills model loading, healthy state, request processing |
≈ EUR 4.727/hour | unknown |
| | |||||
| 1× A100 80 GB VRAM/GPU · fixed | Koyeb A100 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 1.409/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 1.409/hour | ||||
1 geography evidence item
| |||||
| 2× NVIDIA H200 141 GB VRAM/GPU · fixed | Koyeb 2x H200 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 5.284/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 5.284/hour | ||||
1 geography evidence item
| |||||
| 8× A100 80 GB VRAM/GPU · fixed | Koyeb 8x A100 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 11.27/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 11.27/hour | ||||
1 geography evidence item
| |||||
| 4× H100 80 GB VRAM/GPU · fixed | Koyeb 4x H100 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 8.807/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 8.807/hour | ||||
1 geography evidence item
| |||||
| 8× H100 80 GB VRAM/GPU · fixed | Koyeb 8x H100 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 17.61/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 17.61/hour | ||||
1 geography evidence item
| |||||
| 1× A100 80 GB VRAM/GPU · fixed | Koyeb A100 SXM | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 1.893/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 1.893/hour | ||||
1 geography evidence item
| |||||
| 1× NVIDIA RTX PRO 6000 Blackwell 96 GB VRAM/GPU · fixed | Koyeb RTX-PRO-6000 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 1.937/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 1.937/hour | ||||
1 geography evidence item
| |||||
| 4× A100 80 GB VRAM/GPU · fixed | Koyeb 4x A100 SXM | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 7.574/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 7.574/hour | ||||
1 geography evidence item
| |||||
| 4× A100 80 GB VRAM/GPU · fixed | Koyeb 4x A100 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 5.636/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 5.636/hour | ||||
1 geography evidence item
| |||||
| 8× NVIDIA H200 141 GB VRAM/GPU · fixed | Koyeb 8x H200 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 21.14/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 21.14/hour | ||||
1 geography evidence item
| |||||
| 1× H100 80 GB VRAM/GPU · fixed | Koyeb H100 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 2.202/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 2.202/hour | ||||
1 geography evidence item
| |||||
| 2× H100 80 GB VRAM/GPU · fixed | Koyeb 2x H100 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 4.403/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 4.403/hour | ||||
1 geography evidence item
| |||||
| 8× A100 80 GB VRAM/GPU · fixed | Koyeb 8x A100 SXM | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 15.15/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 15.15/hour | ||||
1 geography evidence item
| |||||
| 2× A100 80 GB VRAM/GPU · fixed | Koyeb 2x A100 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 2.818/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 2.818/hour | ||||
1 geography evidence item
| |||||
| 1× NVIDIA H200 141 GB VRAM/GPU · fixed | Koyeb H200 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 2.642/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 2.642/hour | ||||
1 geography evidence item
| |||||
| 2× A100 80 GB VRAM/GPU · fixed | Koyeb 2x A100 SXM | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 3.787/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 3.787/hour | ||||
1 geography evidence item
| |||||
| 1× NVIDIA B200 180 GB VRAM/GPU · fixed | Koyeb B200 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 4.844/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 4.844/hour | ||||
1 geography evidence item
| |||||
| 4× NVIDIA H200 141 GB VRAM/GPU · fixed | Koyeb 4x H200 | Serverless Serverless Compute | always-on always on 1 second increments · bills running |
≈ EUR 10.57/hour | unknown |
| scale-to-zero scale to zero 1 second increments · bills running |
≈ EUR 10.57/hour | ||||
1 geography evidence item
| |||||
| 1× A100 80 GB VRAM/GPU · fixed | Microsoft Azure NVIDIA A100 | Serverless Container Apps Serverless GPU | eastus scale to zero 1 second increments · bills active |
Component price — not a total | EU compute EU-constrained United States · Italy · Sweden |
| swedencentral scale to zero 1 second increments · bills active |
Component price — not a total | ||||
6 geography evidence items
| |||||
| 1× A100 80 GB VRAM/GPU · fixed | Modal Nvidia A100, 80 GB | Serverless Serverless | Broad region scale to zero none minimum · bills running |
Component price — not a total | unknown |
| Narrow region scale to zero none minimum · bills running |
Component price — not a total | ||||
| Standard scale to zero none minimum · bills running |
Component price — not a total | ||||
5 geography evidence items
| |||||
| 4× NVIDIA H200 131.25 GB VRAM/GPU · fixed | OVHcloud h200-4-gpu | Serverless AI Deploy | always-on always on · dedicated 1 minute increments · bills SCALING, RUNNING |
≈ EUR 24.02/hour | unknown |
| scale-to-zero scale to zero · dedicated 1 minute increments · bills SCALING, RUNNING |
≈ EUR 24.02/hour | ||||
| | |||||
| 1× H100 80 GB VRAM/GPU · fixed | OVHcloud h100-1-gpu | Serverless AI Deploy | always-on always on · dedicated 1 minute increments · bills SCALING, RUNNING |
≈ EUR 2.985/hour | unknown |
| scale-to-zero scale to zero · dedicated 1 minute increments · bills SCALING, RUNNING |
≈ EUR 2.985/hour | ||||
| | |||||
| 1× A100 80 GB VRAM/GPU · fixed | OVHcloud a100-1-gpu | Serverless AI Deploy | always-on always on · dedicated 1 minute increments · bills SCALING, RUNNING |
≈ EUR 2.95/hour | unknown |
| scale-to-zero scale to zero · dedicated 1 minute increments · bills SCALING, RUNNING |
≈ EUR 2.95/hour | ||||
| | |||||
| 8× A100 80 GB VRAM/GPU · fixed | Replicate 8x Nvidia A100 (80GB) GPU | Serverless Public Models | Public models request metered · commitment bills request execution time |
≈ EUR 35.51/hour | unknown |
| | |||||
| 1× H100 80 GB VRAM/GPU · fixed | Replicate Nvidia H100 GPU | Serverless Public Models | Public models request metered bills request execution time |
≈ EUR 4.835/hour | unknown |
| | |||||
| 4× A100 80 GB VRAM/GPU · fixed | Replicate 4x Nvidia A100 (80GB) GPU | Serverless Public Models | Public models request metered · commitment bills request execution time |
≈ EUR 17.75/hour | unknown |
| | |||||
| 1× A100 80 GB VRAM/GPU · fixed | Replicate Nvidia A100 (80GB) GPU | Serverless Public Models | Public models request metered bills request execution time |
≈ EUR 4.439/hour | unknown |
| | |||||
| 4× H100 80 GB VRAM/GPU · fixed | Replicate 4x Nvidia H100 GPU | Serverless Public Models | Public models request metered · commitment bills request execution time |
≈ EUR 19.34/hour | unknown |
| | |||||
| 2× H100 80 GB VRAM/GPU · fixed | Replicate 2x Nvidia H100 GPU | Serverless Public Models | Public models request metered · commitment bills request execution time |
≈ EUR 9.67/hour | unknown |
| | |||||
| 8× H100 80 GB VRAM/GPU · fixed | Replicate 8x Nvidia H100 GPU | Serverless Public Models | Public models request metered · commitment bills request execution time |
≈ EUR 38.68/hour | unknown |
| | |||||
| 2× A100 80 GB VRAM/GPU · fixed | Replicate 2x Nvidia A100 (80GB) GPU | Serverless Public Models | Public models request metered bills request execution time |
≈ EUR 8.877/hour | unknown |
| | |||||
| 1× NVIDIA RTX PRO 6000 Blackwell 96 GB VRAM/GPU · fixed | Runpod RTX 6000 Pro | Serverless Serverless | Flex scale to zero 1 second increments · bills initialization, execution, idle timeout |
≈ EUR 3.074/hour | unknown |
| Active workers have discounts available through sales inquiry; no public price is published, so Active variants were omitted. | |||||