Radicalbit AI Monitoring is a self-hosted platform for monitoring production LLM and machine-learning models. It checks numerical and categorical data for anomalies, missing values and outliers, and follows metrics over time. Available model metrics include Precision, Accuracy, Recall, F1, MSE, MAE, Perplexity and Probability. Drift detection offers methods including Kolmogorov-Smirnov, PSI, Wasserstein, Jensen-Shannon, Chi-square, Kullback-Leibler and Hellinger. Users can compare reference and current datasets to assess data quality, model quality and drift. LLM tracing records requests, prompts and tools, then displays sessions, traces and spans with hierarchy, duration and metadata. The v1.4.0 documentation focuses on binary and multiclass classification and regression, with other model types planned. A web UI and Python SDK are provided; the REST API requires PostgreSQL, Kubernetes for Spark metric jobs and distributed storage. Local deployment uses Docker Compose with K3s and supports MinIO or AWS S3. The free, Apache 2.0-licensed plan is listed at no cost.
Who it is for
This platform suits teams responsible for monitoring production classification, regression or LLM applications that can manage a self-hosted deployment. Its API and Python SDK may suit teams integrating monitoring into their own systems.
What is good
- Checks for anomalies, missing values and outliers.
- Includes multiple model metrics and drift methods.
- Compares reference and current datasets.
- Offers a web UI and Python SDK.
- Docker Compose deployment supports MinIO or AWS S3.
What to know first
- Documented scope focuses on classification and regression.
- REST API requires PostgreSQL and distributed storage.
- Spark metric jobs require Kubernetes.
Verdict
Radicalbit AI Monitoring covers data quality, model metrics, drift and LLM traces in a self-hosted package. Confirm that its documented model scope and deployment requirements fit your environment.
Radicalbit AI Monitoring plans and pricing
All plansCompared on machine learning model monitoring software
- Free plan
- Yes
- Drift monitoring
- Yes
- Model performance metrics
- Yes
- Data quality checks
- Yes
- Deployment options
- self-hosted
