{
  "updated": "2026-08-13",
  "refresh_policy": "Verified against vendor pricing pages by the HostFleet data agent. Target cadence: Monday and Thursday.",
  "providers": [
    { "id": "runpod_pods", "name": "RunPod Pods", "billing": "per-second, always-on container (Secure Cloud)", "source": "https://www.runpod.io/pricing" },
    { "id": "runpod_sls", "name": "RunPod Serverless", "billing": "per-second, scale-to-zero workers", "source": "https://www.runpod.io/pricing" },
    { "id": "modal", "name": "Modal", "billing": "per-second, scale-to-zero ($30/mo free credit)", "source": "https://modal.com/pricing" },
    { "id": "fal", "name": "Fal.ai", "billing": "per-hour list price for custom deployments; committed-use discounts advertised", "source": "https://fal.ai/pricing" },
    { "id": "baseten", "name": "Baseten", "billing": "per-minute, dedicated deployments, scale-to-zero", "source": "https://www.baseten.co/pricing/" },
    { "id": "replicate", "name": "Replicate", "billing": "per-second, private model deployments", "source": "https://replicate.com/pricing" },
    { "id": "lambda_cloud", "name": "Lambda Cloud", "billing": "per-hour on-demand GPU instances; prices vary by instance GPU count", "source": "https://lambda.ai/instances" },
    { "id": "coreweave_inference", "name": "CoreWeave Inference", "billing": "per-hour single-GPU inference rate; inference platform customers only (contact account executive)", "source": "https://www.coreweave.com/pricing" },
    { "id": "nebius", "name": "Nebius AI Cloud", "billing": "per-second VM compute, shown as on-demand GPU-hour; H100/H200/B200/B300/RTX PRO rates include prescribed vCPU and RAM, while L40S is the minimum all-in 1-GPU preset", "source": "https://nebius.com/prices" },
    { "id": "hyperstack", "name": "Hyperstack", "billing": "per-minute on-demand GPU VM, shown per GPU-hour; fixed CPU, RAM, root disk, and ephemeral disk are included, while public IPs and shared storage are separate", "source": "https://www.hyperstack.cloud/gpu-pricing" },
    { "id": "novita", "name": "Novita AI", "billing": "per-second on-demand GPU instance, settled hourly; listed one-GPU compute prices include the product's fixed vCPU and RAM, with storage above the free container-disk quota billed separately", "source": "https://novita.ai/gpus-console/explore" },
    { "id": "verda", "name": "Verda (formerly DataCrunch)", "billing": "prepaid pay-as-you-go GPU instances in 10-minute increments, with unused terminated time refunded in the next billing period; listed one-GPU price includes fixed CPU and RAM, while storage is separate", "source": "https://verda.com/pricing" },
    { "id": "salad", "name": "Salad Container Engine", "billing": "per-second managed container instances at Lowest (formerly Batch) priority; GPU rates include selected vCPU and RAM, allocation and image-download time are unbilled, and distributed nodes are interruptible", "source": "https://salad.com/pricing" }
  ],
  "gpus": [
    { "id": "t4", "name": "T4", "vram": "16 GB",
      "prices": {
        "modal": { "hr": 0.59, "raw": "$0.000164/sec" },
        "baseten": { "hr": 0.63, "raw": "$0.01052/min" },
        "replicate": { "hr": 0.81, "raw": "$0.000225/sec" }
      } },
    { "id": "l4", "name": "L4", "vram": "24 GB",
      "prices": {
        "runpod_pods": { "hr": 0.39 },
        "runpod_sls": { "hr": 0.69, "note": "24 GB tier (L4 / A5000 / 3090)" },
        "modal": { "hr": 0.80, "raw": "$0.000222/sec" },
        "baseten": { "hr": 0.85, "raw": "$0.01414/min" }
      } },
    { "id": "a10", "name": "A10 / A10G", "vram": "24 GB",
      "prices": {
        "modal": { "hr": 1.10, "raw": "$0.000306/sec", "note": "A10" },
        "baseten": { "hr": 1.21, "raw": "$0.02012/min", "note": "A10G" },
        "lambda_cloud": { "hr": 1.29, "raw": "$1.29/GPU/hr", "note": "1x A10 instance" }
      } },
    { "id": "rtx4090", "name": "RTX 4090", "vram": "24 GB",
      "prices": {
        "runpod_pods": { "hr": 0.69 },
        "runpod_sls": { "hr": 1.10 },
        "novita": { "hr": 0.33, "raw": "$0.33/GPU/hr", "note": "1x RTX 4090 on-demand product; 16 vCPU, 62 GB RAM, and 60 GB container disk quota" },
        "salad": { "hr": 0.204, "raw": "$0.204/hr", "note": "Lowest (formerly Batch) priority managed container; public configuration shows 4 vCPU and 8 GB RAM included; interruptible distributed node, not a dedicated VM" }
      } },
    { "id": "rtx5090", "name": "RTX 5090", "vram": "32 GB",
      "prices": {
        "runpod_pods": { "hr": 0.99 },
        "runpod_sls": { "hr": 1.58 },
        "novita": { "hr": 0.72, "raw": "$0.72/GPU/hr", "note": "1x RTX 5090 high-frequency on-demand product; 24 vCPU, 58 GB RAM, and 60 GB container disk quota" },
        "salad": { "hr": 0.294, "raw": "$0.294/hr", "note": "Lowest (formerly Batch) priority managed container; public configuration shows 4 vCPU and 8 GB RAM included; interruptible distributed node, not a dedicated VM" }
      } },
    { "id": "a6000", "name": "RTX A6000 / A40", "vram": "48 GB",
      "prices": {
        "runpod_pods": { "hr": 0.44, "note": "A40 $0.44, RTX A6000 $0.53" },
        "runpod_sls": { "hr": 1.22, "note": "A6000 / A40 tier" },
        "lambda_cloud": { "hr": 1.09, "raw": "$1.09/GPU/hr", "note": "1x NVIDIA A6000 instance" },
        "hyperstack": { "hr": 0.50, "raw": "$0.50/GPU/hr", "note": "1x RTX A6000 PCIe flavor; 28 CPU, 58 GB RAM, and local storage included; CANADA-1" },
        "verda": { "hr": 0.61, "raw": "$0.6100/hr", "note": "1x RTX A6000 48 GB on-demand instance; 10 CPU and 60 GB RAM included; storage separate" }
      } },
    { "id": "l40s", "name": "L40S", "vram": "48 GB",
      "prices": {
        "runpod_pods": { "hr": 0.99 },
        "runpod_sls": { "hr": 1.75, "note": "48 GB tier (L40 / L40S / 6000 Ada)" },
        "modal": { "hr": 1.95, "raw": "$0.000542/sec" },
        "replicate": { "hr": 3.51, "raw": "$0.000975/sec" },
        "coreweave_inference": { "hr": 2.25, "raw": "$2.25/GPU/hr", "note": "Inference platform customers only; public 8x L40S on-demand instance is $18.00/hr" },
        "nebius": { "hr": 1.55, "raw": "$1.35 GPU + 8x$0.012 vCPU + 32x$0.0032 GiB RAM = $1.5484/hr", "note": "Minimum all-in Intel 1gpu-8vcpu-32gb preset, rounded to the vendor's published from-$1.55 rate; eu-north1" },
        "novita": { "hr": 0.55, "raw": "$0.55/GPU/hr", "note": "1x L40S on-demand product; 22 vCPU, 125 GB RAM, and 60 GB container disk quota" },
        "verda": { "hr": 1.37, "raw": "$1.37/hr", "note": "1x L40S 48 GB on-demand instance; 20 CPU and 60 GB RAM included; storage separate" }
      } },
    { "id": "rtxpro6000", "name": "RTX PRO 6000", "vram": "96 GB",
      "prices": {
        "runpod_pods": { "hr": 1.99 },
        "runpod_sls": { "hr": 3.49 },
        "modal": { "hr": 3.03, "raw": "$0.000842/sec" },
        "fal": { "hr": 2.99, "note": "as low as $1.10 committed" },
        "coreweave_inference": { "hr": 2.50, "raw": "$2.50/GPU/hr", "note": "RTX PRO 6000 Blackwell Server Edition; inference platform customers only; public 8x on-demand instance is $20.00/hr" },
        "nebius": { "hr": 1.80, "raw": "$1.80/GPU/hr", "note": "RTX PRO 6000 Server Edition; unified 1gpu-24vcpu-218gb VM rate; us-central1" },
        "hyperstack": { "hr": 1.85, "raw": "$1.85/GPU/hr", "note": "RTX PRO 6000 Server Edition; 1x flavor has 28 CPU, 225 GB RAM, and local storage included; CANADA-1" },
        "verda": { "hr": 1.89, "raw": "$1.89/hr", "note": "1x RTX PRO 6000 96 GB on-demand instance; 30 CPU and 90 GB RAM included; non-confidential-compute flavor; storage separate" }
      } },
    { "id": "a100_40", "name": "A100 40 GB", "vram": "40 GB",
      "prices": {
        "modal": { "hr": 2.10, "raw": "$0.000583/sec" },
        "lambda_cloud": { "hr": 1.99, "raw": "$1.99/GPU/hr", "note": "1x A100 PCIe or A100 SXM; 40 GB" },
        "verda": { "hr": 1.29, "raw": "$1.29/hr", "note": "1x A100 40 GB SXM4 on-demand instance; 22 CPU and 120 GB RAM included; storage separate" }
      } },
    { "id": "a100_80", "name": "A100 80 GB", "vram": "80 GB",
      "prices": {
        "runpod_pods": { "hr": 1.39, "note": "PCIe $1.39, SXM $1.49" },
        "runpod_sls": { "hr": 2.72 },
        "modal": { "hr": 2.50, "raw": "$0.000694/sec" },
        "baseten": { "hr": 4.00, "raw": "$0.06667/min" },
        "replicate": { "hr": 5.04, "raw": "$0.001400/sec" },
        "coreweave_inference": { "hr": 2.70, "raw": "$2.70/GPU/hr", "note": "Inference platform customers only; public 8x A100 80 GB on-demand instance is $21.60/hr" },
        "hyperstack": { "hr": 1.35, "raw": "$1.35/GPU/hr", "note": "1x A100 80 GB PCIe flavor; 28 CPU, 120 GB RAM, and local storage included; CANADA-1" },
        "novita": { "hr": 1.60, "raw": "$1.60/GPU/hr", "note": "1x A100 80 GB SXM on-demand product; 14 vCPU, 240 GB RAM, and 60 GB container disk quota" },
        "verda": { "hr": 1.79, "raw": "$1.79/hr", "note": "1x A100 80 GB SXM4 on-demand instance; 22 CPU and 120 GB RAM included; storage separate" }
      } },
    { "id": "h100", "name": "H100 80 GB", "vram": "80 GB",
      "prices": {
        "runpod_pods": { "hr": 2.89, "note": "PCIe $2.89, SXM $2.99, NVL $3.19" },
        "runpod_sls": { "hr": 4.55 },
        "modal": { "hr": 3.95, "raw": "$0.001097/sec" },
        "fal": { "hr": 4.50, "note": "as low as $1.89 committed" },
        "baseten": { "hr": 6.50, "raw": "$0.10833/min" },
        "replicate": { "hr": 5.49, "raw": "$0.001525/sec" },
        "lambda_cloud": { "hr": 3.29, "raw": "$3.29/GPU/hr", "note": "1x H100 PCIe; H100 SXM is $4.29/GPU/hr" },
        "coreweave_inference": { "hr": 6.16, "raw": "$6.16/GPU/hr", "note": "Inference platform customers only; public 8x HGX H100 on-demand instance is $49.24/hr" },
        "nebius": { "hr": 3.85, "raw": "$3.85/GPU/hr", "note": "H100 SXM/NVLink; unified 1gpu-16vcpu-200gb VM rate; eu-north1" },
        "hyperstack": { "hr": 2.50, "raw": "$2.50/GPU/hr", "note": "1x H100 80 GB PCIe flavor; 28 CPU, 180 GB RAM, and local storage included; CANADA-1" },
        "novita": { "hr": 3.39, "raw": "$3.39/GPU/hr", "note": "1x H100 80 GB SXM on-demand product; 16 vCPU, 128 GB RAM, and 60 GB container disk quota" },
        "verda": { "hr": 3.25, "raw": "$3.25/hr", "note": "1x H100 80 GB SXM5 on-demand instance; selected 30 CPU and 120 GB RAM configuration; another 32 CPU/185 GB RAM configuration has the same price; storage separate" }
      } },
    { "id": "h200", "name": "H200", "vram": "141 GB",
      "prices": {
        "runpod_pods": { "hr": 4.39 },
        "runpod_sls": { "hr": 5.93 },
        "modal": { "hr": 4.54, "raw": "$0.001261/sec" },
        "fal": { "hr": 4.50, "note": "as low as $2.10 committed" },
        "coreweave_inference": { "hr": 6.31, "raw": "$6.31/GPU/hr", "note": "Inference platform customers only; public 8x HGX H200 on-demand instance is $50.44/hr" },
        "nebius": { "hr": 4.50, "raw": "$4.50/GPU/hr", "note": "H200 SXM/NVLink; unified 1gpu-16vcpu-200gb VM rate; public regions include eu-north1, eu-west1, and us-central1" },
        "hyperstack": { "hr": 3.99, "raw": "$3.99/GPU/hr; 8x flavor = $31.92/hr", "note": "H200 SXM5 is cataloged only as an 8x flavor with 176 CPU, 1800 GB RAM, and local storage included; CANADA-1" },
        "verda": { "hr": 4.00, "raw": "$4.00/hr", "note": "1x H200 141 GB SXM5 on-demand instance; 44 CPU and 182 GB RAM included; storage separate" }
      } },
    { "id": "b200", "name": "B200", "vram": "180 GB",
      "prices": {
        "runpod_pods": { "hr": 5.89 },
        "runpod_sls": { "hr": 8.64 },
        "modal": { "hr": 6.25, "raw": "$0.001736/sec" },
        "fal": { "hr": 6.25, "note": "as low as $3.49 committed" },
        "baseten": { "hr": 9.98, "raw": "$0.16633/min" },
        "lambda_cloud": { "hr": 6.99, "raw": "$6.99/GPU/hr", "note": "1x B200 SXM6 instance" },
        "coreweave_inference": { "hr": 8.60, "raw": "$8.60/GPU/hr", "note": "Inference platform customers only; public 8x HGX B200 on-demand instance is $68.80/hr" },
        "nebius": { "hr": 7.15, "raw": "$7.15/GPU/hr", "note": "B200 SXM/NVLink; unified 1gpu-20vcpu-224gb VM rate; us-central1 or me-west1" },
        "hyperstack": { "hr": 6.00, "raw": "$6.00/GPU/hr; 8x flavor = $48.00/hr", "note": "B200 SXM6 is cataloged only as an 8x flavor with 252 CPU, 2048 GB RAM, and local storage included; vendor specifies 192 GB VRAM; CANADA-1" },
        "verda": { "hr": 6.11, "raw": "$6.11/hr", "note": "1x B200 180 GB SXM6 on-demand instance; 30 CPU and 170 GB RAM included; storage separate" }
      } },
    { "id": "b300", "name": "B300", "vram": "288 GB",
      "prices": {
        "runpod_pods": { "hr": 7.39 },
        "runpod_sls": { "hr": 9.98 },
        "modal": { "hr": 7.10, "raw": "$0.001972/sec" },
        "fal": { "hr": 8.50, "note": "as low as $4.49 committed" },
        "nebius": { "hr": 7.85, "raw": "$7.85/GPU/hr", "note": "B300 SXM/NVLink; unified 1gpu-24vcpu-346gb VM rate; uk-south1 and private eu-west2" },
        "verda": { "hr": 7.50, "raw": "$7.50/hr", "note": "1x B300 268 GB SXM6 on-demand instance; 30 CPU and 255 GB RAM included; vendor lists 268 GB VRAM; storage separate" }
      } }
  ]
}