Just like machine learning models, pricing can be opaque, especially in the high-growth MLOps space. As a mission-driven company helping teams achieve responsible AI by ensuring model outcomes are fair and trustworthy, we believe that our pricing should be in the same vein — grounded in transparency.
Over the past two years, we have held hundreds of pricing discussions with MLOps buyers, giving us deep insight into the purchase considerations for an MLOps lifecycle solution. With this feedback, today we’re excited to announce a new pricing model featuring bespoke plans and metrics.
The objective of our new pricing is two fold:
- Make pricing simple and frictionless so that our customers only pay for what they need
- Meet customers where they are on their AI journey and grow with them as they scale AI across their organizations
Here are a few key benefits to our new pricing model:
- New Pricing Plans - We are introducing the new ‘Lite’ and ‘Standard’ pricing plans to cater to teams in different stages of the AI journey.
The ‘Lite’ plan comes with core model monitoring and explainable AI capabilities to help small to mid-sized ML teams get started at an affordable price point.
The ‘Standard’ plan additionally features enterprise-level capabilities including security compliance, AI fairness assessment, advanced explainability, and Virtual Private Cloud (VPC) deployments for established ML teams. With the ‘Standard’ plan, you can also purchase add-on capabilities and services, including GPU accelerated explanations and expert AI support.
- A la Carte Pricing - Depending on use cases, customers may have different visibility needs for their ML models. Regulated use cases or complex ML models have a more immediate need for explainability, while high volume unregulated use cases might have a stronger need for monitoring. We call these ‘explainability-first’ or ‘monitoring-first’ customers. They may eventually expand to use all the explainability, monitoring, and fairness capabilities in Fiddler as they mature in their AI journey. To enable this expansion, the new pricing model decouples monitoring, explainability and fairness - so you only pay for the capability you use.
- Unified Pricing Across Deployments - In the purchase process, many customers are initially undecided about managed or virtual cloud deployment and hope to decide after a product trial. With Fiddler’s new pricing, you can get price comparisons for the same usage across any deployment model to do an apples to apples comparison. Our managed cloud pricing is significantly cheaper than VPC even though it includes the underlying cloud costs, given the high overhead of supporting VPC installations.
- Usage based - We have incorporated elements of consumption-based pricing in our methodology and with clear pricing for each metric. Our pricing will, however, still require an annual commitment. Teams are already familiar with consumption-based pricing from software solutions and data clouds like Snowflake, AWS, and GCP. Now we are making it easy for them to calculate and only pay for what they need from an AI Observability solution.
But how do we calculate the pricing for monitoring and explainability? To do this, we use four simple metrics.
- Data Ingested - The size of all data to be uploaded.
We looked at comparable APM solutions in order to use a pricing model our customers are already comfortable with. Fiddler is now the first MPM solution to adopt data ingestion as its primary pricing metric! This has the added benefit of being closer to true consumption-based pricing, since each prediction from a model with fewer features is cheaper than one with many features. The data ingested metric includes the size of all model predictions and metadata, baselines, and model artifacts uploaded into the system.
- Models - The number of models to be monitored or explained.
- Explanations - The number of model inferences to be explained.
- Data Retention - The number of months raw data will be retained. This is only applicable to managed cloud deployments.
To illustrate this, below is an example of how we would calculate the price of a ‘Lite’ plan for a monitoring-first customer:
Learn more about the capabilities included in each plan and request a pricing estimate by visiting our new Pricing page.