OpenObserve

OpenObserve is an observability platform for logs, metrics and traces that stores
telemetry in object storage rather than on expensive local disk, which is where its cost claims come
from — the project positions itself at roughly 4-8x lower spend than Datadog for comparable volume.
It is OpenTelemetry-native, so instrumentation is the standard OTel SDK rather than a proprietary
agent.

Why it shows up in an AI directory. Alongside conventional application monitoring, OpenObserve
ships LLM and agent observability: traces for model calls across OpenAI, Anthropic, Gemini, Bedrock,
Mistral, Ollama, Groq, Together and vLLM, and for orchestration frameworks including LangChain,
LangGraph, LlamaIndex, CrewAI, Semantic Kernel, DSPy, Pydantic AI, the Vercel AI SDK, the Claude
Agent SDK and OpenAI Agents. Gateways and proxies (LiteLLM, Portkey, OpenRouter, Kong, Vercel AI
Gateway) and eval tooling (Ragas, Promptfoo) are covered too, and there is an MCP server so an agent
can query your own telemetry.

Ingest surface. Data arrives over OTLP, or from Prometheus, Vector, Fluent Bit, Fluentd,
Filebeat, Kinesis Firehose, syslog and plain curl. Pre-built pipelines cover Kubernetes, AWS (EC2,
Lambda, ECS, RDS, CloudFront, VPC flow logs, WAF), GCP, Azure, Cloudflare, Vercel, nginx, the usual
databases, Kafka, Okta and Microsoft 365.

What you get in the product. Full-text and SQL search over logs, real-time analytics and
dashboards, data parsing and enrichment pipelines, front-end/RUM monitoring, alerting, and role-based
access with an audit trail.

Licensing and pricing. The open-source edition is AGPL-3.0 and free forever to self-host. A
self-hosted Enterprise build — SSO, RBAC, audit trail, data redaction — is free up to 50 GB of
ingestion per day, with paid support above that. Cloud Professional is usage-priced at $0.50 per GB
ingested and $0.01 per GB queried with no per-seat fee (15 months metric retention, 30 days for logs
and traces) and a 14-day trial; Cloud Enterprise is custom-priced and supports infinite retention in
your own bucket.