Langfuse vs. Raindrop

This guide outlines the key differences between Langfuse and Raindrop to help engineering teams choose the right AI engineering platform. All Raindrop facts were checked against public Raindrop pages in October 2026.

TL;DR

• Open-source core, free to self-host. Langfuse's core tracing, evaluation, prompt management, and APIs are MIT-licensed and available on Cloud or self-hosted; additional Enterprise features require a license. Self-hosting is free on every tier, with Enterprise SSO and organization RBAC included. Raindrop's monitoring platform is a commercial hosted product: Workshop, the MIT-licensed part, is a local debugger, and self-hosting is advertised as a beta.
• About 65% lower on the illustrated workload, with no seat fees. 500,000 user turns at 12 Langfuse units each is $621 on Langfuse Pro where Raindrop Pro is $1,799: a $299 base plus $0.003 per event. Both figures are subscription plus telemetry usage; model inference and add-ons are separate. Langfuse Pro includes 3 years of history and unlimited users; Raindrop Pro includes unlimited retention.
• Data through APIs and exports, or SQL when self-hosting. Langfuse runs on ClickHouse, open source since 2016; read it through the Observations and Metrics APIs, scheduled exports to your bucket on Pro with the Teams add-on, Enterprise, and self-hosted, or SQL on your own ClickHouse. Raindrop data export starts on Pro, and warehouse exports are Enterprise.
• The improvement loop in one project. Langfuse runs prompt and model experiments in the UI and complete agents through the SDK, compares results to a baseline, and keeps annotation queues, prompt labels, and CI gates in the same project. Raindrop's hosted experiments compare traffic already logged and do not run or replay the agent.
• Same-day switch. Point the SDK or OpenTelemetry exporter at Langfuse and import dataset cases via CSV, SDK, or API. Most teams see first traces in Langfuse the same day.

Choose Langfuse if you want to own the platform: an open-source core for production traces, datasets, experiments, annotation, and prompts, on Cloud or self-hosted including air-gapped, on ClickHouse you can query and export. One meter covers unlimited users, and on the worked example it is about 65% lower than Raindrop Pro at list prices. Proven at 21 of the Fortune 50.

Choose Raindrop if you are already running it and the differences on this page do not justify a switch. You accept a commercial hosted platform with self-hosting in beta, a $299 base plus per-event pricing, a Triage message allowance on Pro, warehouse export on Enterprise, and hosted experiments that compare logged traffic rather than running the agent.

---

Open source and deployment

Langfuse's core tracing, evaluation, prompt management, and APIs are MIT-licensed and available on Cloud or self-hosted; additional Enterprise features require a license. Self-hosting runs the UI, login, and data in your environment on every tier, including air-gapped operation, and Enterprise SSO and organization-level RBAC ship in the free distribution. Langfuse Cloud is covered by SOC 2 Type II and ISO 27001 audits, is GDPR compliant, and signs a BAA from the Pro plan in the HIPAA region. Langfuse is used by 21 of the Fortune 50 and 129 of the Fortune 500.

Raindrop's monitoring platform is a hosted, paid product. Workshop, the MIT-licensed part, is a local debugger, and self-hosting is advertised as a beta. Its security docs place event storage in AWS with United States subprocessors.

---

Storage and data access

ClickHouse has been open source since June 15, 2016 (Apache 2.0) and is today the most popular open-source analytical database, with 2,000+ contributors. Langfuse Cloud and self-hosted run the same ClickHouse architecture, so you can move between them without locking into a closed store.

Langfuse works directly with ClickHouse's core team as part of ClickHouse. Every LLM call, tool execution, and agent step is written once to a wide, immutable observations table (details). High-volume reads go through the Observations and Metrics APIs, scheduled exports (Pro with the Teams add-on, Enterprise, and self-hosted) land JSONL or Parquet in your own S3, GCS, or Azure bucket, and self-hosters query ClickHouse in SQL.

Raindrop's store is vendor-operated. Reads go through its Query API and MCP; data export is a Pro feature on the plans page, and the pricing cards put Snowflake and BigQuery exports on Enterprise.

---

Pricing

Raindrop bills events. The pricing card defines one event as a user message and agent response, including tool calls and sub-agents, or one background agent run. Langfuse bills units: one trace, one observation, or one score. A single Raindrop event can contain many Langfuse observations, so the cheaper bill depends on trace shape.

The table and example below cover the software subscription and telemetry usage only. Model inference for judges or Triage, privacy add-ons, and export add-ons are separate on both sides.

Masking and PII Guard are different approaches, not the same feature at two prices: Langfuse masks in your SDK with rules you define before data is sent; Raindrop's PII Guard detects and redacts on its side after ingestion.

Worked example

500,000 user turns/month · 10 observations per turn · 1 score per turn. Each turn is one Raindrop event; in Langfuse it is 1 trace + 10 observations + 1 score = 12 units.

Public list prices. About 65% lower on Langfuse Pro in this illustrated workload: $621 versus $1,799 per month.

• Langfuse Pro: 6,000,000 units. 100k included, 900k at $8/100k ($72), 5,000,000 at $7/100k ($350). Base $199. 3-year data access.
• Raindrop Pro: 500,000 events at $0.003. Base $299. Unlimited retention.
• Excluded on both sides: model inference for judges or Triage, privacy add-ons (Raindrop PII Guard), and export add-ons (the Langfuse Teams add-on for scheduled bucket export).

Raindrop bills the turn; Langfuse bills every observation and score inside it. A lean turn is cheaper on Langfuse Pro; a turn with dozens of model calls, tool spans, and scores can be cheaper on Raindrop Pro.

See Langfuse pricing, start free, or talk to us about Enterprise volumes.

---

The improvement loop

Langfuse runs the loop in one project. Add a production observation to a dataset, then run an experiment: prompt and model experiments execute prompts and models in the UI, and a complete agent runs through an SDK experiment in your repository with your existing mocks and test environment, with results compared to a baseline in Langfuse. Route outputs to annotation queues. Prompt labels promote a version without a redeploy, and the version can carry model and tool config that your application reads at runtime. LLM-as-a-judge and code evaluators score live observations and experiment runs; judges use a model connection you choose. In production, Pulse surfaces outliers, alerts fire on thresholds you define, and Halo investigates traces and creates datasets, prompts, and dashboards on request; it is on every Cloud plan and in public beta for self-hosted.

Raindrop's docs describe the product as monitoring for silent agent failures: issue detection, custom signals, a Triage agent, datasets, and A/B tests. Its hosted experiments compare cohorts of logged events and, in Raindrop's words, "don't run or replay your agent"; Workshop and Simulations provide separate replay and testing workflows.

---

Which should you choose

Choose Langfuse if you want an open-source platform that works across your stack: OpenTelemetry instrumentation for any framework, free self-hosting with SSO and organization RBAC included, prompt experiments in the UI and full-agent experiments through the SDK against a curated dataset, annotation queues and prompt labels on the same traces, data in ClickHouse you can query and export, and unit pricing with no seat fees.

Choose Raindrop if you are already running it and the differences above do not justify a switch.

---

Why teams move away from Raindrop

The same reasons recur in migration conversations.

• Event pricing with a $299 floor. Raindrop Pro starts at $299 and charges $0.003 per event, so the bill scales with every user turn. On the worked example that is $1,799 on Raindrop Pro versus $621 on Langfuse Pro, which meters units with no seat fees and includes 3 years of history. The gap shrinks when one event contains dozens of observations.
• The platform you buy is hosted; the open-source artifact is a local debugger. Raindrop's Workshop is free and local, but the monitoring platform teams pay for is the hosted plan, with self-hosting in beta and warehouse export on Enterprise. Langfuse self-hosts its open-source core, UI, login, and evaluators included, on ClickHouse you can query.
• Hosted experiments do not run the agent. Raindrop's hosted experiments compare logged traffic and do not run or replay the agent. Teams that need a candidate scored against a curated set run prompt experiments in the Langfuse UI or a full-agent experiment through the SDK, then promote a prompt label from the result with a CI gate.
• Triage is metered in messages. Raindrop Pro includes a Triage allowance of 500 messages a month, with unlimited on Enterprise. Compare it separately with Halo's plan-based usage limits; evaluator execution and threshold alerts are different workloads and are not a substitute for either.

---

Why teams choose Langfuse

A pattern repeats across production adopters: start where governance requires, often self-hosted, then grow into the platform.

• Canva runs a multi-agent support system for a product with 250 million monthly active users. LLM-as-a-judge evaluators score the systems across 15–20 metrics, and domain experts build evaluators without an engineering handoff. They self-hosted first, then moved to Cloud.
• SumUp rolled AI support to 35+ markets and deflected almost 50% of conversations, cutting external BPO cost by 30%. They started self-hosted for the proof of concept, then moved to Langfuse Cloud.
• Merck runs about 80 of 300+ GenAI use cases on Langfuse, with 200+ people building on it. They self-host for data sovereignty and provision projects via API.

---

Switching from Raindrop

Move live instrumentation and recreate the durable assets. History stays readable in Raindrop for the plan's retention window. Most teams see first traces in Langfuse the same day.

• Instrumentation: initialize the Langfuse SDK or send OpenTelemetry to the OTLP endpoint. Run both exporters during the overlap.
• Datasets, signals, and prompts: export a Raindrop dataset as CSV and import it through the Langfuse CSV uploader, SDK, or API. Recreate judges and prompts in Langfuse. A Raindrop signal is a detector on logged events; the matching Langfuse object is an evaluator or a score.
• Trace history: stays in Raindrop. Pro's unlimited retention means the archive can remain there while new data lands in Langfuse.

Talk to us if you want help planning the overlap.

---

Start free: Cloud or self-host

Langfuse Cloud Hobby includes 50k units/month, no credit card. Or self-host the MIT stack. The example project is populated before you instrument anything.

---

FAQ

---

---