Skip to content
Sign inRequest access

Products

GPU CloudAMD MI325X and NVIDIA H200 by the hour or on reservation.
Dedicated ClustersPreviewPrivate, interconnected GPU clusters for sustained workloads.
SandboxesPreviewIsolated, short-lived environments for developers and AI agents.
Serverless GPUPreviewGPU containers that scale with requests, down to zero.
Serverless InferenceOpen models behind one OpenAI-compatible endpoint.
Dedicated EndpointsReserved serving capacity for one model, billed per GPU-hour.
PlaygroundSend a prompt to any model your workspace can reach.
Model DeploymentsPreviewImmutable deployments with instant promote and rollback.
Fine-tuningLoRA fine-tunes on the open models in our catalog.
Model TrainingPreviewFoundation-model training runs, managed end to end.
EvaluationsPreviewBenchmark suites and model comparisons you can rerun.
DatasetsManaged and synthetic training data, versioned and ready to use.
Data ManagementPreviewDatasets, connections and storage, with lineage across the platform.
PipelinesPreviewVersioned DAGs that automate data, training and deployment.
Data RoomsPreviewSecure document sharing with guests, NDAs and an audit trail.
NokuvaPreviewThe AI-native design editor.
TavocPreviewA Rust-native code platform: IDE and desktop apps.
FoltracComing soonMulti-cloud deployment with autonomous operations.
W0rkTreeComing soonVersion control for agent fleets.
Identity & AccessMembers, roles, policies, SSO and audit for every workspace.
Billing & CostUsage, invoices, credits and budgets in one place.
Logs & MonitoringRequest logs, latency and error rates across services.
Project CanvasPreviewSee and change a project's resources on one live canvas.

Pages

Home/
Platform/platform
All products/products
Pricing/pricing
GPU pricing/pricing#compute
Inference services/services/inference
Model training services/services/model-training
Protocol design/services/protocol-design
Research partnerships/services/research-partnerships
Protocols/protocols
Datasets/datasets
Documentationdocs.legionedge.ai
Changelog/changelog
News/news
Research/research
Blog/blog
Articles/articles
Events/events
Startup Program/programs/startup
Ambassador Program/programs/ambassador
Partner Program/programs/partner
Event Program/programs/events
Company/company
Careers/careers
Partners/partners
Contact/contact
Statusstatus.legionedge.ai
Terms of Service/legal/terms
Privacy Policy/legal/privacy

Dedicated Endpoints

Available

Your model, on GPUs nobody else is using.

Deploy a catalog model or your own fine-tune on GPUs reserved for you, with predictable latency and no shared queue.

Request accessView pricing
Pricing
$4.79 / GPU-hr
Layer
Inference
Deploy
Catalog models, Fine-tune checkpoints, Hugging Face repos, Uploaded weights, NVIDIA H200, AMD Instinct MI325X

Why Dedicated Endpoints

Predictable latency, no shared queue.

A dedicated endpoint serves one model on capacity reserved for you. You choose the hardware, the regions and the replica bounds, it scales between them, and traffic can stay off the public internet.

  • Any model

    A catalog model, a fine-tune checkpoint, a Hugging Face repository and revision, or weights uploaded from a dataset bucket.

  • Autoscaling

    Replicas scale between the minimum and maximum you set, against a concurrency target, with scale-to-zero if you want it.

  • Private networking

    Keep traffic off the public internet, and allow only the keys and IP ranges you list.

How it works

Reserved serving capacity for one model, billed per GPU-hour.

  1. Choose the model

    Start from the catalog, a fine-tune checkpoint, a Hugging Face repo and revision, or weights you uploaded.

  2. Fit the hardware

    Pick H200 or MI325X. The wizard checks the model fits in GPU memory at your quantization and shows the hourly price.

  3. Set the bounds

    Choose regions, minimum and maximum replicas, a concurrency target and the serving engine, and turn on scale-to-zero if idle hours matter.

  4. Serve and scale

    Watch replicas come up with their region, GPU and status, then let autoscaling move between the bounds you set.

In the console

Everything in one place.

Dedicated Endpoints lives in the LegionEdge cloud console, next to every other service in the workspace, under one account and one identity.

  • Hardware

    GPU cards with their memory, checked against the model before anything is provisioned.

    • NVIDIA H200 and AMD Instinct MI325X
    • Memory fit check for model size and quantization
    • A hardware change ships as a new deployment

  • Scaling

    Replicas between the bounds you set, in the regions you choose.

    • Minimum and maximum replicas
    • A target concurrency per replica
    • Optional scale-to-zero, with a cold start

  • Serving

    The engine and configuration your model runs with, frozen per deployment.

    • vLLM, SGLang or TGI
    • Quantization chosen per deployment
    • Live replicas with per-replica logs

  • Networking and access

    The controls a production endpoint needs, set per project.

    • Private networking, off the public internet
    • Allowed keys and IP ranges, with per-key limits
    • Custom domains, verified with a TXT record

Pricing

$4.79 / GPU-hr

Dedicated endpoints are billed per GPU-hour for the GPUs their replicas run on.

  • Reserved terms lower the hourly rate on MI325X capacity.
  • Scale-to-zero is optional, so idle capacity can be released.
  • The wizard shows a cost estimate before you deploy.
Full pricing

GPU rates

GPU rates in USD per GPU-hour, on demand and by reservation term
GPUMemoryOn demand1-month6-month12-month
AMD Instinct MI325X256 GB HBM3E$4.79$4.31$3.83$3.35
NVIDIA H200141 GB HBM3e$4.79On request

Regions

Compute regions
RegionLocationHardwareStatus
US Centralus-central-ordChicago, ILAMD MI325XLive
US Southus-south-1Dallas, TXNVIDIA H200Live
US Eastus-east-1Ashburn, VANVIDIA H200Planned

Related products

Pairs well with Dedicated Endpoints

Dedicated Endpoints

Capacity that answers to one model.

Deploy a catalog model or your own fine-tune on GPUs reserved for you.

Request accessTalk to us