pricing

Customize your computing needs at significantly lower costs

Spot -20%
Up to 20% for short-term projects
$2.99
NVIDIA H100
on-demand
  • $2.99 NVIDIA SXM H100
  • $2.69 NVIDIA H100 PCIe
  • $1.89 NVIDIA A100 80GB NVLINK
  • $1.79 NVIDIA A100 80GB PCIe
  • $1.49 NVIDIA A100 40GB NVLINK
  • And many more
  • Storage
  • $0.06 per GB Block Storage NVMe
  • $0.04 per GB Block Storage HDD
  • $0.03 per GB Object Storage NVMe
  • $0.03 per GB Object Storage HDD
  • $0.06 per GB EFS File System Storage NVMe
  • Feature Included
    Reserved
    For long-term projects, please send us a personal request
    $2.59

    6 months renting price for H100
  • $NVIDIA HGX H100
  • NVIDIA H100 PCIe
  • NVIDIA A100 80GB NVLINK
  • NVIDIA A100 80GB PCIe
  • NVIDIA 40GB PCIe
  • NVIDIA A40
  • Tesla V100 NVLINK
  • Storage
  • Block Storage NVMe
  • Block Storage HDD
  • Object Storage NVMe
  • Object Storage HDD
  • EFS File System Storage
  • Enterprise
    For a more customized solution, please get in touch
  • Full Time Support
  • Flexible Solutions
  • Custom agreements for specific needs
  • API Pricing

    Pricing is based on the time or volume it takes to run your model. Some models are billed per token, others per image, per second of audio, or video. These examples represent only a portion of our offerings – you’ll find detailed pricing for all models on the model page.

    Language Models

    Model Input Output
    deepseek-ai/deepseek-r1 $3.750 / 1M tokens $10.000 / 1M tokens
    deepseek-ai/deepseek-v3 $1.450 / 1M tokens $1.450 / 1M tokens
    deepmind/gemma-3-4b-it $0.120 / 1M tokens $0.300 / 1M tokens
    deepmind/gemma-3-12b-it $0.250 / 1M tokens $0.700 / 1M tokens
    deepmind/gemma-3-27b-it $0.480 / 1M tokens $1.200 / 1M tokens
    anthropic/claude-3.5-haiku $1.000 / 1M tokens $5.000 / 1M tokens
    anthropic/claude-3.5-sonnet $3.750 / 1M tokens $18.750 / 1M tokens
    anthropic/claude-3.7-sonnet $3.000 / 1M tokens $15.000 / 1M tokens
    ibm-granite/granite-3.0-2b-instruct $0.030 / 1M tokens $0.250 / 1M tokens
    meta/llama-2-13b $0.100 / 1M tokens $0.500 / 1M tokens
    meta/llama-2-70b $0.650 / 1M tokens $2.750 / 1M tokens
    meta/llama-4-maverick-instruct $0.250 / 1M tokens $0.950 / 1M tokens
    meta/llama-4-scout-instruct $0.170 / 1M tokens $0.650 / 1M tokens
    meta/meta-llama-3.1-405b-instruct $9.500 / 1M tokens $9.500 / 1M tokens
    meta/meta-llama-3-70b $0.650 / 1M tokens $2.750 / 1M tokens
    meta/meta-llama-3-70b-instruct $0.650 / 1M tokens $2.750 / 1M tokens
    meta/meta-llama-3-8b $0.050 / 1M tokens $0.250 / 1M tokens
    meta/meta-llama-3-8b-instruct $0.050 / 1M tokens $0.250 / 1M tokens
    mistralai/mistral-7b-instruct-v0.2 $0.050 / 1M tokens $0.250 / 1M tokens
    mistralai/mistral-7b-v0.1 $0.050 / 1M tokens $0.250 / 1M tokens

    Batch Processing Discount

    For large volume processing, we offer a 50% discount on most models using the Snowcell Batch API.

    Image Models

    Model Output
    stability-ai/stable-diffusion-3.5-large $0.055 / image
    stability-ai/stable-diffusion-3 $0.025 / image
    google/imagen-3 $0.040 / image
    google/imagen-3-fast $0.015 / image
    recraft-ai/recraft-20b $0.012 / image
    ideogram-ai/ideogram-v2 $0.070 / image
    black-forest-labs/flux-1.1-pro $0.030 / image
    luma/photon $0.020 / image

    Audio Models

    Model Output
    playht/play-dialog $0.060 / minute
    minimax/music-01 $0.035 / audio file
    openai/whisper-large-v3 $0.005 / minute
    microsoft/sesame-csm-1b $0.004 / minute

    Video Models

    Model Output
    luma/ray $0.450 / video
    google/veo-2 $0.500 / second

    Enterprise or custom deployments?

    Get private models, SLAs, dedicated GPUs, and usage-based pricing for high-throughput workloads.

    Contact Sales

    Subscription Pricing

    Subscription is for Snowcell Chat service. Our Pro Plan offers a straightforward per-seat subscription, while the Enterprise Plan is tailored for larger organizations with custom requirements.

    Pro Plan

    Designed for individual developers and small teams, this plan includes dedicated GPU access, priority technical support, and usage-based overages after your included compute hours.

    $18 / seat/month
    • Wide range of models
    • Access to technical support
    • Included compute hours with usage-based overages
    • Access to pre-hosted inference models
    Get Started

    Enterprise Plan

    Our Enterprise Plan offers a fully customized solution with private deployments, SLAs, dedicated account management, and priority escalated support.

    Custom Pricing
    • Custom deployments and private hosting
    • Service Level Agreements (SLAs)
    • Priority escalated support
    • Volume discounts and flexible billing
    Contact Sales
    Dedicated Endpoints

    Isolated, Customizable GPU Endpoints

    Deploy pre-selected models or your custom fine-tuned versions on dedicated GPU endpoints with predictable, per-minute billing. Start or stop endpoints at your convenience using our web UI, API, or CLI—all while enjoying isolated performance.

    Hardware Type Price/Minute Price/Hour
    1x RTX-6000 48GB $0.025 $1.49
    1x L40 48GB $0.025 $1.49
    1x L40S 48GB $0.035 $2.10
    1x A100 PCIe 80GB $0.040 $2.40
    1x A100 SXM 40GB $0.040 $2.40
    1x A100 SXM 80GB $0.043 $2.56
    1x H100 80GB $0.056 $3.36
    1x H200 141GB $0.083 $4.99
    Fine-Tuning GPU Pods

    Pay Only for the Time You Fine-Tune

    Fine-tuning jobs are executed by deploying dedicated GPU pods that run for the duration of your training process.

    When you submit a fine-tuning job, our orchestrator provisions a GPU pod configured with the hardware you need—such as NVIDIA A100 or H100. This pod is exclusively reserved for your fine-tuning task, ensuring consistent performance and full isolation. You’re billed only for the time the pod is active, making it a transparent and predictable way to manage fine-tuning costs.

    Frequent questions and answers

    Fully automated, fully supported

    Self-serve or work with a representative, we're here to help.

    How is Snowcell cheaper than legacy providers?

    At Snowcell, our mission is clear: we aim to deliver unmatched performance at the most competitive price point available. With the industry’s most diverse selection of GPUs, we ensure you have the ideal resources for your unique workload. Our infrastructure is engineered for both speed and adaptability, enabling peak performance and cost-effectiveness. And when it comes to support, we’re personally committed to optimizing your infrastructure for maximum efficiency. With Snowcell, you’re not just choosing a service, you’re choosing a partner dedicated to empowering your success.

    Do you offer spot instances?

    Yes, we do. Our on-demand pricing mirrors the flexibility of interruptible instances, but with the assurance of uninterrupted service. Our clients enjoy the best of both worlds—scalable solutions with reliable access, so once you spin up an instance on SnowCell Cloud, it remains yours until you're ready to wind down, ensuring your workflows proceed without a hitch.

    Do you offer discounts?

    At Snowcell, you can dive into flexible on-demand cloud solutions with no strings attached. For consistent workloads, we provide substantial savings through Reserved Instances and Bulk Credits, tailored to your usage. Chat with our Sales team to unlock discounts that boost your business's efficiency and savings.

    Are there any hidden costs?

    We don't believe in charging you for every GB transferred. We've eliminated fees for regional transfers and workstation data in most scenarios. If your bandwidth needs are extensive, we'll collaborate closely with you to understand your requirements and secure flat rates, ensuring your large-scale cloud operations remain cost-effective. If you are unsure about the pricing plan and/or other costs, just drop us an email at [email protected].

    Get Started

    Launch your servers today at minimal costs.