One cockpit for your
entire GPU fleet.
Everything the agent sees and does, laid out the way an operator needs it. Scroll through the stations.
The whole fleet on one screen.
Live earnings rate, accepted rentals, utilization, and agent status. The numbers that matter, refreshed while you watch.
- Live earnings rate and potential per day, side by side
- Needs-attention queue flags idle GPUs and cheaper competitors
- Every machine's floor, strategy, and rental state at a glance
Know the market before you price it.
A live read on demand across thousands of similar rigs, and exactly where your fleet sits inside it.
- Demand snapshot built from thousands of comparable rigs
- Your fleet vs. market utilization, measured head to head
- See where each price lands between market P10 and P90
Pick the pricing play, per machine.
Market Match by default. Advanced modes for hosts who want to compete on performance or protect premium inventory.
- Market Match follows what similar GPUs actually rent for
- Performance Match prices against your true DLPerf peers
- Premium Multi-GPU defends large-rental inventory
Every decision shows its work.
No black box. Each price change is logged with the market data, peer band, and guardrail check behind it.
- Full reasoning attached to every update and hold
- Price Guardrail records approvals and floor clamps
- Filter by machine, marketplace, and event type
Rentals and revenue, quantified.
Utilization, rental prices, and revenue estimates over time, so you know what the agent is earning you.
- Rented vs. idle GPU-hours, day by day
- Rental density heatmap by day of week and hour
- GPU-hour-weighted average rental price across the fleet
Weekends reprice themselves.
GPU demand follows weekly cycles. Set day-of-week floor overrides once and the agent adapts as the week turns.
- Per-day price-floor overrides, automatically applied
- Normal floors restored when the override day ends
- Copy one machine's week to the rest of the fleet
Margins after the electric bill.
Your kWh rate, each GPU's power draw, and real utilization, combined into the true cost behind every rental.
- Daily, weekly, and monthly cost projections per machine
- Modeled from GPU TDP, system overhead, and 24h utilization
- Know the margin behind every accepted rate
Host vitals, streamed live.
GPU workload, temperature, CPU, memory, disk, and network on one synchronized timeline, straight from the host.
- Up to 1,440 samples per host per day
- Stacked metric bands with synchronized hover
- Host configuration, uptime, and agent health included
Up and running in 3 minutes.
No software on your rigs, no config files. Connect, set a floor, and hand the market to the agent.
Connect your marketplace
Paste your API key. It's encrypted at rest and your machines are discovered automatically. Nothing to install on your rigs.
Set your floor
Pick a price floor per machine and keep Market Match as the default strategy. The agent never prices below your floor.
Flip on the autopilot
From here on the agent scans the market every 3 minutes and repositions your listings: overnight, weekends, holidays.
One agent. Every marketplace.
Live where you host today, expanding to where you'll host next. Every integration gets the same 3-minute pricing loop.
RunPod·
Nosana·
FluxEdge·
AkashRun a marketplace and want your hosts on autopilot? Talk to us
Built by a host.
Run by a fleet of them.
- 0+
- GPUs under management
- 0
- pricing decisions / system / month
- 0s
- from market move to reprice
from single rigs to datacenter racks
one every 3 minutes, day and night
your listing reacts within a single cycle
Simple for one rig. Serious for a fleet.
From single-marketplace automation to full multi-marketplace orchestration.
GPU Autopilot Agent
Intelligent pricing for GPU marketplace hosts
- 24/7 automated price monitoring
- Market Match default with VAST strategy controls
- Real-time market analytics (GPU IQ)
- Price floor, strategy, and bid offset controls
- Interruptible (min-bid) pricing
- Day-of-week price scheduler
- Power cost tracking and analysis
- Multi-platform notifications
- Auto-discovery of machines
- Complete activity and audit trail
- Cancel anytime
Autopilot Manager
Multi-Marketplace Orchestration Platform
Unified GPU orchestration across multiple marketplaces simultaneously. Powered by proprietary vGPU framework technology built in-house, the Manager dynamically allocates GPU capacity, automates cross-platform maintenance windows, and optimizes pricing across every listing.
Everything in Agent, plus:
- Unified orchestration across multiple GPU marketplaces
- Proprietary vGPU framework technology
- Dynamic GPU allocation and capacity management
- On-premise agent with full hardware telemetry
- Automated cross-marketplace maintenance windows
- White-glove deployment and dedicated consultation
- Cross-platform earnings aggregation dashboard
- Real-time system control and GPU passthrough management
Questions, answered.
Everything hosts usually ask before switching on the agent.
Once you connect a supported marketplace API key and select your machines, our agent monitors the market 24/7 and automatically adjusts your GPU listing prices based on supply, demand, rented prices, and competitor pricing.
Yes. Your marketplace API keys are encrypted at rest and only used to discover machines, read marketplace status, and update prices for the machines you enable.
A system is a single machine or node on a supported marketplace, regardless of how many GPUs it has. A 1x RTX 4090 and an 8x A100 both count as one system at $5/mo each.
Absolutely. There are no contracts or commitments. Cancel through the billing portal and your subscription ends at the current billing period. Turning agents off only pauses automation and does not cancel billing.
No install is required for pricing automation. GPU Autopilot runs as a cloud service and communicates through marketplace APIs. The optional Telemetry page has a separate host agent if you want hardware, thermal, and network monitoring.
Analyzer AI is a coming-soon, research-only assistant for GPU market questions, hardware comparisons, 500.farm chart interpretation, and context you choose to provide. It will not change fleet settings or prices.
Yes. Two-factor authentication is optional and can be enabled from Settings with Google Authenticator, Microsoft Authenticator, 1Password, or another TOTP-compatible authenticator app.
The agent targets the rented median price: what renters actually pay for your GPU type. If a cheaper competitor is listed below that median, the agent undercuts them to win sort position. Your configured floor and platform safeguards are always respected. When you're already the cheapest listing, the agent raises your price toward the next competitor to maximize revenue.
The agent checks market conditions and adjusts pricing every few minutes. You can configure each machine with its own floor, bid offset, region filter, strategy, and rental protection behavior through the dashboard.
Still have questions? Ask us on Discord or Telegram