Comparisons anchored on Langfuse, the open-core LLM observability platform that can be self-hosted.
Open-core LLM observability you can self-host freely against a commercial platform whose self-hosted option is an Enterprise add-on, comparing licensing, the shared ClickHouse and PostgreSQL footprint and the trade-off for teams with data residency rules.
Whether to self-host Langfuse with Docker Compose on one VM or on Kubernetes with the Langfuse Helm chart. Covers the six components every deployment runs (web, worker, PostgreSQL, ClickHouse, Redis or Valkey, object storage), what Langfuse says Compose lacks, what the chart bundles and what to move to managed services for production.
Whether a self-hosted Langfuse deployment needs an Enterprise license key: what the MIT edition already includes without limits, the RBAC, retention, audit log, masking and SCIM features the key unlocks, how Enterprise is now bundled with ClickHouse commercial plans, and what happens if the key is removed.
Whether to send LLM traces to Langfuse Cloud or keep them on your own Langfuse stack: unit-based cloud pricing and data access windows by plan, US, EU and JP regions, which compliance reports and SSO features sit in which tier, and the PostgreSQL, ClickHouse, Redis and blob storage you run when self-hosting.
SDK and OpenTelemetry tracing against a proxy-first gateway that logs requests by changing the base URL. Compares self-hosted architecture (both store analytics in ClickHouse), tracing, prompt management and evaluations, licensing, and what Helicone's maintenance mode after Mintlify's acquisition in March 2026 means for teams choosing or leaving it.
Two self-hostable LLM observability platforms compared: Langfuse and Comet's Apache-2.0 Opik. Covers tracing, evaluations and prompt management, which features each keeps for paid editions, the databases each needs (Opik runs on ClickHouse, MySQL and Redis) and the effort to operate them.
Open-source Langfuse against Arize Phoenix, which is source-available under the Elastic License 2.0. Compares license limits on hosting as a service, OpenTelemetry and OpenInference tracing, evaluations and datasets, authentication options and SQLite or PostgreSQL storage.
The three self-hosted LLM tracing and evaluation tools teams shortlist, compared on license type, feature gating between free and paid editions, storage architecture, instrumentation standards and the infrastructure each needs.
LLM tracing and evaluation in Langfuse against MLflow's OpenTelemetry-based tracing, evaluation and prompt registry: licenses and paid features, what each needs to self-host (ClickHouse, PostgreSQL and Redis versus an MLflow server with a SQL backend), and whether an existing MLflow estate makes a second tool unnecessary.
Tell us what you're working on, whether it's a deployment, an audit, a security test or a cyber range. You'll speak with an engineer who can help you scope it.