Start building for free. Upgrade for higher usage limits, dedicated support, and enterprise-grade hosting options.
Free
No credit card required
Get started.png)
10K events per month
Up to 5 users
Single workspace
30d data retention
Full evaluation, observability, and prompt management suite
Let's chat
Ideal for large organizations
Book a demoCustom usage limits
Unlimited users and workspaces
Choose between multi-tenant SaaS, dedicated SaaS, or self-hosting
Custom SSO & SAML
Dedicated support, SLA, and team trainings
.png)
.png){kind=logo}
Div Garg
Co-Founder
%20(1).png)
Rex Harris
Head of AI/ML
%20(1).png)
.png)
Cristian Pinto
CTO
An event refers to a single trace span or metric-label combination sent to our API as OTLP or JSON. It captures any relevant data from your system, including all context fields generated by your application's instrumentation. In simple terms:-
Number of Events = Number of Trace Spans + Number of Metrics
HoneyHive supports 2 primary types of evaluations:
Automated Evaluations
These are functions—either code-based or using LLM-as-a-judge—that automatically score your sessions, agents, or spans. They generate measurable scores and provide explanations for their assessments. Common examples include Context Relevance, Answer Faithfulness, ROUGE-L, Tool Use Accuracy, etc. HoneyHive provides dozens of standard evaluators out-of-the-box, and you can also define custom evaluators tailored to your specific needs.
Human Evaluators
We strongly recommend a hybrid evaluation approach that combines automation with human oversight. This helps you account for evaluator bias and ensures alignment with your domain experts' standards. HoneyHive lets you create custom scoring rubrics and annotation queues that domain experts can use to manually grade outputs, ensuring your metrics truly reflect what matters for your use case.
All data is secure and encrypted at rest and in transit. We are SOC-2 Type II, GDPR, and HIPAA compliant, conduct regular penetration tests via 3rd-party auditors, and provide flexible hosting solutions, including self-hosting, to meet your security and compliance needs. Learn more about our platform architecture here.
Yes, you can self-host HoneyHive on the Enterprise plan. We support self-hosting across AWS, Azure, and Google Cloud via Kubernetes HELM charts, and can provide additional support for on-premise deployments. Contact us to learn more.
You can log traces using our SDKs, or async using our batch ingestion APIs.
We offer SDKs in Python and Typescript with native OpenTelemetry support, and provide automatic instrumentation for 50+ popular libraries like LangChain, LangGraph, AWS Strands, Google ADK, and OpenAI Agents SDK, among others.
For users using other languages, you can send your OpenTelemetry traces to our OTEL collector or manually instrument your application using our APIs.
Yes, we do offer startup discounts for companies with less than $5M of total funding raised. Contact us to learn more.