Cloud GPU hosts and clients that focus on ROI and meeting monthly KPIs with partners and investors, require insights that enable them to spec and price their servers with the least expensive hardware to get the highest possible revenue and utilization. Here is what you need to know:

• What type of GPU and how many should I put in my server and how should that sever be specified?
• How many minimum GPUs should I set in my offer?
• Is 1 better than 4? Do 4 take longer to get hired, but lock in utilization? How to decide?
• How do I set my initial price for highest utilization?
• How long should a lower price be offered before lowering it again?
• Is a higher price at a lower utilization better than a lower price with higher utilization?

We developed this Cloud Analytics Web application to answer these questions, to gain insights from metric breakdowns by number of GPUs in a machine and/or offer, standard deviation of rented offer prices beyond just percentiles, correlations of hardware specs to rented prices, and historical reference and analysis of all that data.

We ingest every offer and machine configuration from Vast's data provider 500Farm. We use this to determine GPU cloud prices, utilization, contract durations, and implications of supporting hardware on prices. We also use that data and rolled up analysis to train AI that drives the Woven Auto Listing Agent.

Choosing a GPU should be based on your budget and bus plan’s financial model. It should take into account GPU, hardware, electrical, data, hvac, and co-location costs.

The offer analytics dashboard enables you to investigate offers on Vast.ai for specific types of GPUs and number per offer. You can review live and historical utilization, rented and vacant prices, offer flow, verification periods, machine specs and utilization of rented and vacant servers.

The Machine Utilization dashboard allows you to investigate utilization and earnings for specific GPUs and quantities in a machine. You can also review historically rented and vacant prices along with utilization on Vast.ai.

Use the machine configuration ranges to price correlations to see what market rented prices are on Vast.ai for the configuration you are planning. This helps you not over or under specify your machine for the highest earnings at the lowest cost.

Use the offer price distribution panel and the machine to market price efficiencies to see what price other hosts are listing at and where you are recommended to list.

Offers volume can be monitored using the rented and vacant historical offers to see if users are renting, and you need to adjust your price, or they are not, and you just need to be patient or change the minimum number of GPUs in an offer if utilization is better.

Auto sort placement on Vast.ai is key to getting rented quickly. The Woven Listing Agent increases auto sort placement on Vast by automatically setting prices every 60 seconds to a suggested price range. The prices are derived from the price efficiency analytics that compares your machine's price to all like rented and configured peers.

Adjusts your prices as soon as high volume clients (whales) arrive and prices shoot up, then lowers as oon as they depart. Never miss another boom again.

Helps you set prices relative to your verification state for maximum earnings. Adjusts rental and bid prices from unverified to verified as soon as it happens.

Allows you to control minimum prices if the suggested price is below your minimum threshold. Also give you complete control of minimum GPU, discount rate, storage and Inet prices, as well as the contract end date.

Deploy the Woven Auto-Listing Agent on any machine that has the Vast CLI app installed and your API key configured for it. WE DO NOT NEED YOUR API KEY

Woven’s revenue and utilization strategy is focused on optimizing a machine’s offer prices to maximize auto sort position in the Vast Rental UI. This default view of rigs for clients is based on Vast’s score metric.

Auto sort is your Vast score. It is a calculated using a secret formula, but we have very strong data that show its most heavily impacted by price, internet speed, reliability, and DLPerf (listed in order of maximum impact). Once online, a machine’s DLPerf will not change in any significant way and reliability should only rise, so price is the only variable to tune that can impact auto sort placement.

As mentioned earlier, the Woven Listing Agent uses suggested price levels the Woven Offer Analytics engine derives for every single machine from similar rented offers in the market. These offers have key hardware metrics and verification states within range or matching that of the machine’s price performance being measured.

The suggested prices are tested from highest to lowest, never exceeding a minimum set by the user or via our automation, until we get the offer as high up the auto sort as we can. If we are at the top, we make sure the price isn’t too low by testing our score against the previous offer in the auto sort behind us. When vacant we are changing that price every 1 minute to increase visibility.

Machines on Vast have a life cycle that traverses verification state and climbs a reliability scale. These are important because verification states have completely different market prices and utilization. Offers with reliability under 90% do not show up in the default view of the Vast Rental UI, regardless of their Vast score. Verification of a machine takes approximately 2-4 weeks, and reliability climbs fastest when listed and rented.

Contact us via Discord or email for more info.

The Listing Agent Client requires an Inconel subscription.

The Inconel sub includes Vast RTL & Events:

Listing Agent - Price Automation ($USD/month):

GPU listing automation licenses can be purchased in-app once and Inconel Analytics subscriber.

Examples GPU Servers:

Contact us via Discord or email for more info.