Pricing
Find the plan that’s right for your team
Pro
For small and establishing teams
$50/mo per model
(Save 30% with annual billing)
(Save 30% with annual billing)
- Up to 2 models or apps
Enterprise
For teams with advanced needs or global scale
- Contact us to customize your plan
- Custom number of models or apps
LLM Use Case
Logs (Spans)
Storage
ML & Computer Vision Use Case
Features Per Model
Production Predictions
Latent Ground Truth / Updates
Key Features
LLM Tracing
AI-Powered SQL Filter Builder
Sessions
Offline (Code-Based) LLM Evaluations
Online Evaluations - LLM as a Judge
Online Evaluations - Python/TS Code Evaluator
AI-Powered Evals Builder
Prompt Playground
AI-Powered Prompt Optimizer
Prompt Hub
Dynamic Datasets
Annotations
Experiments
AI-Powered RAG Debugger
Core Capabilities
Embeddings Analysis
Embeddings Similarity Search
AI-Powered Search
Performance Tracing
AI-Powered Analytics
Dashboards
AI-Powered Dashboard Builder
Monitors
Alerting Integrations
Bias Tracing
Explainability
Custom Metrics
AI-Powered Custom Metrics builder
Data Export
API Access
Security & Controls
Organizations
Workspaces
History Lookback / Retention
Social Sign On
SAML SSO
RBAC
On-premise deployment option
Compliance reporting
HIPAA reporting
Support
Support
SLA
Customer success manager
Customer success engineering
Pro
For small and establishing teams
$50/mo per model
(Save 30% with annual billing)
(Save 30% with annual billing)
- Up to 2 models or apps
10K per month
10GB
200
1M per month
14-day lookback period
3 per app
5GB
1
1
6 months
Community, email
Business hours
Enterprise
For teams with advanced needs or global scale
- Contact us to customize your plan
- Custom number of models or apps
Custom volume
Custom volume
1000+
Custom volume
Customizable lookback
Unlimited
Unlimited
Unlimited
Unlimited
13 months (optional increase)
Slack, email, phone
3hr response window
Option
Marketplace & Travel Company Standardizes Observability Across AI/ML Teams
"We found that the platform offered great exploratory analysis and model debugging capabilities, and during the POC it was able to reliably detect model issues."