Free

$0 forever

Everything you need for local GPU monitoring.

Local GPU monitoring
Apple Silicon, NVIDIA, AMD, Intel
15 color themes (Monokai, Dracula, Nord, etc.)
7 view modes (Grid, Detail, List, Predict, Compare, Topology, Fleet)
OOM prediction & leak detection
Per-process GPU attribution (%UTL, %CPU, %MEM)
Prometheus /metrics endpoint
OpenTelemetry export
SQLite data persistence
Keyboard-driven interface
Colorblind-safe themes
No telemetry, no data collection

Download Now

Pro

Coming Soon

$29 /month

For teams monitoring multiple machines.

Everything in Free
Fleet SSH monitoring (up to 20 machines)
Bastion/jump host support
Email alerts
Slack & Discord notifications
Webhook integrations
Web dashboard (coming)
Multi-machine comparison
Priority email support
14-day free trial

Join Waitlist

Enterprise

Coming Soon

Custom

For large-scale GPU fleet management.

Everything in Pro
Cloud API integration (AWS EC2, Azure VMs, GCP Compute)
Agent-based fleet monitoring (50+ GPU nodes)
RBAC & SSO (SAML, OIDC)
GPU scheduling & orchestration
Audit logs
Custom alert policies & escalation
PagerDuty/OpsGenie integration
SLA (99.9% monitoring uptime)
Dedicated support engineer
Custom integrations

Contact Us

COMPARE

What's in each tier?

Every tier includes full local GPU monitoring. Pro adds fleet visibility. Enterprise adds cloud-scale automation.

Feature	Free	Pro PRO
Core Monitoring
Local GPU monitoring
Apple Silicon + NVIDIA + AMD + Intel
15 themes, 7 view modes
OOM prediction & leak detection
Per-process GPU attribution
Prometheus /metrics endpoint
Pro Features
Fleet monitoring (SSH, up to 20 machines)	—
Email, Slack & Discord alerts	—
Memory bandwidth monitoring	—
NVLink bandwidth & error tracking	—
ECC error monitoring	—
Thermal throttle duration tracking	—
Energy consumption tracking	—
Web dashboard	—
Priority support	—
Enterprise Features
Cloud API auto-discovery (AWS/Azure/GCP)	—	—
Agent-based fleet monitoring (unlimited)	—	—
K8s GPU pod monitoring	—	—
Tensor core utilization	—	—
SM/EU occupancy profiling	—	—
MIG partition monitoring	—	—
RBAC & SSO (SAML, OIDC)	—	—
GPU scheduling & orchestration	—	—
Custom alert policies & escalation	—	—
PagerDuty/OpsGenie integration	—	—
Audit logs & SLA	—	—
Dedicated support engineer	—	—

ROI

Why Pro pays for itself

GPU compute is expensive. Wasted GPU time is more expensive. gpulse Pro catches problems before they cost you thousands.

$30

per idle H100 hour

A single H100 sitting idle on cloud costs ~$30/hr. gpulse Pro at $29/mo catches just one hour of waste and you've already hit ROI.

$17,280

to restart a crashed 72h training run

A 72-hour training run on 8x H100s that crashes to OOM costs $17,280 to restart. OOM prediction preventing one crash pays for 50 years of Pro.

1 hour

to pay back your first month

Pro monitors up to 20 machines. Catch one idle GPU for one hour on any of them and the subscription has already paid for itself.

0 min

setup time

No agents to deploy, no config files to write. SSH into your cloud VMs (AWS, Azure, GCP) or on-prem servers and start monitoring immediately.

FAQ

Pricing questions

Why is the free tier free?

We believe every developer should have access to GPU monitoring. The free tier is fully functional — it includes all local monitoring features, OOM prediction, 15 themes, and every view mode. No artificial limitations.

When will Pro be available?

We're currently in beta. Join the waitlist to be notified when Pro launches. Early waitlist members will get extended trial access.

Can I try Pro features?

Pro will include a 14-day free trial when available. No credit card required to start the trial.

What payment methods do you accept?

Credit card and annual invoicing for Pro. Enterprise customers can arrange custom billing including purchase orders and wire transfers.

Is there a student/academic discount?

Yes. Contact info@gpulse.ai with your .edu email for academic pricing.

Still have questions?

Contact us at info@gpulse.ai