Explore ModelCloud
PlatformSolutionsPricingDocsCompanyContact
MODEL HOSTING & INFERENCE CLOUD

You bring the model.
We run it.

Run your own open-weights and fine-tuned models on a managed GPU fleet — without operating a single server. No per-token markup, no walled garden.

Get started ↗Explore the platform
Your weights →Managed GPU fleet→ Endpoint

Deployment flow · conceptual illustration, not a live fleet view.

Trusted by platform teams in fintech, healthcare, robotics, legal, media, and e-commerce

01 / ONE FLEET

Three products.
One GPU fleet.

Each product stands alone or composes with the others — from a single serving endpoint to a compliance-ready dedicated fleet.

GA

Inference

Serve any open-weights or fine-tuned model behind one OpenAI-compatible endpoint.

  • Autoscale replicas from zero to demand
  • Warm pools to cut cold starts
  • FP8 / INT8 / INT4 quantization
View inference →
Beta

Training

Fine-tune and train across many GPUs with checkpointing and deterministic resume.

  • Distributed multi-GPU jobs
  • Spot capacity with auto-reschedule
  • Resume from any checkpoint
View training →
GA

Observability

Latency, throughput, token accounting, and cost attribution per model and deployment.

  • Per-model latency & throughput
  • Token and cost accounting
  • Export logs and traces to your stack
View observability →
Plus, Fleet — dedicated and reserved capacity with VPC peering, quotas, and documented compliance posture (SOC 2 Type I report available under NDA; HIPAA posture on dedicated fleets), with GDPR-compliant data handling. View fleet →
02 / DEPLOYMENT

From weights to
a running endpoint.

Bring your model, pick a GPU, and ModelCloud stands up the endpoint, autoscaling, and telemetry — so you deploy in minutes instead of operating servers.

01 / YOUR MODEL

Bring the weights.

Open-weights · fine-tuned

02 / BUILD & OPTIMIZE

Prepare to deploy.

Containerize · quantize

03 / MANAGED FLEET

Pick a GPU.

Replicas scale with load · autoscale to zero when idle

04 / ENDPOINT

Connect your app.

OpenAI-compatible · one base URL · your API key

Deployment quickstart ↗

Replicas follow load.

Autoscale to zero when idle. Cold start on the next request.

autoscale · replicas follow load, to zero when idle
03 / WORKLOADS

Run models the way your industry needs them

The same fleet, configured for the constraint that matters to you — latency, isolation, compliance, or cost.

Financial services

Sub-second fraud scoring with VPC isolation and multi-region failover.

Healthcare & life sciences

HIPAA posture, private connectivity, and traceable model versioning.

Robotics & logistics

Shift-aware autoscaling and per-line cost attribution for vision models.

Legal & compliance

Bulk document review on spot capacity, kept inside a VPC.

Media & gaming

Burst-ready autoscaling and warm pools for launch-day traffic.

E-commerce

High-concurrency search and recommendations with per-request tracing.
observability · metrics per model & deployment

Conceptual illustration · no operational readings.

scheduling · jobs → managed GPU slots

Conceptual illustration · no operational readings.

04 / CAPACITY & COST

Pay for the GPUs you run

Start free, scale per GPU-hour, or commit to reserved and dedicated capacity at a discount. No per-token markup on your own models.

billing · you pay for GPU-hours you run, not tokens

Starter

For evaluation and side projects.

$0 / mo
no credit card required
  • 40 GPU-hours / month
  • 2 concurrent models
  • Shared capacity
  • Autoscaling to zero
  • Community support
Start free
MOST USED

Production

For serving models in production.

$1.90 / GPU-hr
+ $49/mo platform fee
  • Autoscaling to zero
  • Warm pools (cold-start)
  • Priority queue
  • FP8 / INT8 / INT4 quantization
  • 99.9% SLA, email support
Deploy a model

Scale

For committed, steady-state capacity.

Reserved
committed-use discount
  • Reserved GPU capacity
  • Discounted GPU-hour rate
  • Warm pools always on
  • Multi-region serving
  • 99.95% SLA
Talk to sales

Enterprise

For regulated, high-throughput workloads.

Custom
annual agreement
  • Dedicated GPU fleet
  • VPC peering
  • SSO and audit logs
  • Dedicated support
Talk to us
Representative per-GPU-hour pricingbilled to the second
AcceleratorMemoryBest forOn-demandReserved
L424 GBembeddings, small models$0.70 / hr$0.42 / hr
A10G24 GB7–8B models, high concurrency$1.10 / hr$0.66 / hr
L40S48 GB13–30B models, best price-per-token$1.90 / hr$1.14 / hr
A100 · 40 GB40 GB13–30B models$2.20 / hr$1.32 / hr
A100 · 80 GB80 GB30–70B models, tensor-parallel$3.50 / hr$2.10 / hr
H100 · 80 GB80 GB70B+ models, lowest latency$5.10 / hr$3.06 / hr

Rates are representative for on-demand managed inference. Reserved pricing locks a fixed rate against a minimum commit. Spot capacity is available at a further discount for batch, offline, and evaluation workloads that can tolerate interruption.

Full pricing details →
05 / COMPANY

Built by people who ran production inference

Dana Whitfield

Dana Whitfield

Co-founder & CEO
Marcus Bell

Marcus Bell

Co-founder & CTO
Priya Nair

Priya Nair

Head of Engineering
Kenji Watanabe

Kenji Watanabe

Head of ML
Amara Diallo

Amara Diallo

Head of Product
Lucas Meyer

Lucas Meyer

Head of Infrastructure

Bring your model. We'll run it.

Start on the free tier, move to reserved or dedicated capacity as your traffic grows — and keep your weights yours the whole way.