Both trace agents, run experiments and score production traffic, and both were acquired in 2026. Langfuse is the open-source option you can run yourself at no cost; Galileo, now Splunk Agent Observability, is the paid platform with guardrails on model and tool inputs and outputs during a run. Licence and run-time checks pull in opposite directions here, so decide which matters more.[3][6]
Langfuse is an MIT-licensed platform you can self-host for free, while Galileo is a proprietary platform sold through Splunk that adds guardrails acting during a run. Both changed owners in 2026: ClickHouse acquired Langfuse in January, and Cisco completed its acquisition of Galileo in May, selling it as Splunk Agent Observability since 7 August. Langfuse has a free hosted tier and a $29 a month plan; Splunk lists annual, span-based pricing and no free tier.
| Decision factor | Galileo | Langfuse |
|---|---|---|
| Ownership | Cisco, completed May 2026; sold as Splunk Agent Observability since 7 August 2026 | ClickHouse, January 2026; roadmap and licence unchanged, per Langfuse |
| Licence | Proprietary platform; SDKs and Agent Control are Apache-2.0 | MIT, except add-ons in the ee folders |
| Self-hosting | On-premises, including air-gapped (commercial) | Free, via Docker or Kubernetes |
| Run-time guardrails | Yes: Agent Control checks model and tool inputs and outputs during a run | None; evaluation runs after the fact |
| Prompt tooling | Prompts and a playground; version control not documented | Versioning, label-based deployment, prompts linked to traces, playground |
| OpenTelemetry | Accepts OTLP; OpenInference instrumentors in Python only | OTLP over HTTP; gRPC not supported |
| Free tier | None listed for Agent Observability | Hobby: 50k units a month, 30 days of data; self-hosting free |
| Paid pricing | From $100 a month per 1.2 million spans a year, billed annually | Core: $29 a month, 100k units |
| Capability | Galileo | Langfuse | Prefactor |
|---|---|---|---|
| Observe | |||
| LLM call tracing | ✓ | ✓ | ✓ |
| Multi-step agent tracing (tool calls, nested spans) | ✓ | ✓ | ✓ |
| Cost tracking | ✓ | ✓ | ✓ |
| OpenTelemetry support | ✓ | ✓ | ✓ |
| Evaluate | |||
| Offline evals (datasets, experiments) | ✓ | ✓ | ✓ |
| Online evals on production traffic | ✓ | ✓ | ✓ |
| LLM-as-a-judge scoring | ✓ | ✓ | ✓ |
| Code-based evaluators | ✓ | ✓ | ✓ |
| Human review and annotation | ✓ | ✓ | ✓ |
| Prompt versioning and playground | Partial | ✓ | — |
| Act on findings | |||
| Runtime controls on live agent actions | ✓ | — | ✓ |
| Gateway or proxy in the request path | — | — | — |
| Licence and deployment | |||
| Open source | Partial | ✓ | Partial |
| Self-hosting | ✓ | ✓ | Partial |
Langfuse announced on 16 January 2026 that ClickHouse had acquired it, and said its roadmap and licence would not change and that self-hosting would stay a first-class path. It has kept shipping under the same name: rebuilt experiments in April, code evaluators in May, a 1.0 command-line tool in August and evaluator backfills in September.
Cisco completed its acquisition of Galileo in May 2026, and since 7 August 2026 the product has been sold as Splunk Agent Observability inside Splunk Observability Cloud, with a separate documentation site for customers who joined after that date. The product continues to ship, but it is now bought through Splunk, with annual billing.
One near-term change applies to Langfuse Cloud users: it becomes v4-only on 16 November 2026, which may require migration work for existing projects.
Langfuse publishes its platform under the MIT licence, apart from add-on features in licence-keyed ee folders, and documents free self-hosting on Docker or Kubernetes backed by Postgres, ClickHouse, Redis and object storage. The self-hosted version includes online evaluator rules that score live production traffic.
Galileo's platform is proprietary. It can run on-premises, including air-gapped, as a commercial deployment, or as SaaS in Splunk Observability Cloud. Its Python and TypeScript SDKs and Agent Control are open source under Apache-2.0, and Agent Control can be self-hosted on its own. If a free deployment inside your network is the requirement, Langfuse meets it and Galileo does not.
Hosted options differ too. Langfuse Cloud runs in EU, US, Japan and HIPAA regions, starting on the free Hobby tier. Galileo runs as SaaS within Splunk Observability Cloud, and customers onboarded before 7 August 2026 still use the legacy Galileo Cloud.
Galileo's Agent Control evaluates model and tool inputs and outputs during agent execution and can block prompt injection, personal data leaks and harmful content using rules set in one place, without agent code changes. It replaced the deprecated Protect product in June 2026. Approval workflows for actions are not documented.
Langfuse does not block or enforce anything on live requests. Its documentation says run-time protection comes from separate libraries and that Langfuse evaluates those measures afterwards; threshold alerts to Slack, GitHub Actions or webhooks notify rather than block.
On evaluation itself, both offer LLM judges, code evaluators, human annotation and online scoring with sampling. Galileo adds Luna-2 small models as lower-cost judges, though the Splunk-hosted edition supports only five preset Luna evaluators and no custom code evaluators. Langfuse adds evaluator backfills over up to six months of past data and a CI action that fails a build below a score threshold.
For tracing, Langfuse captures nested observations and sessions, with an agent graph view since July 2026. Galileo logs sessions, traces and spans, added A2A multi-agent tracing in May 2026 and offers an Agent Flow metric for agent paths. Both list integrations for LangGraph, CrewAI, the OpenAI Agents SDK, Google ADK, Strands, Mastra and Pydantic AI, which reduces the work of moving between them.
As of 20 October 2026, Langfuse charges per unit ingested: Hobby is free for 50k units a month, Core is $29 a month for 100k units then $8 per 100k, Pro is $199 a month and Enterprise $2,499 a month or custom. Self-hosting has no subscription but adds the infrastructure you run. Splunk lists Agent Observability from $100 a month per 1.2 million spans a year, billed annually, with a Luna token allowance and no free tier listed.
Retention rises with the Langfuse plan: Hobby keeps 30 days of data, Core 90 days and Pro three years, with a Teams add-on at $300 a month. Splunk lists extended data retention as an option without a published price.
See Splunk's pricing and Langfuse's pricing and model your own volumes; the units differ, so headline prices do not compare directly.
We sell the layer this section describes. Read it with that in mind.
Stated plainly, since Prefactor publishes this page: Prefactor is an agent evaluation platform that scores each step of a run and acts on what it finds through runtime policies on tool calls and data access, approval holds for risky actions, and a kill switch for a run, an agent or a team. Galileo's Agent Control is the closer of these two, since it blocks content during a run; Langfuse evaluates afterwards. Neither documents approval holds. Prefactor has no prompt playground or prompt registry and its platform is hosted, so if free self-hosting or prompt versioning is what you need, Langfuse is the better pick. See the full comparison of 15 platforms.
Reviewed against public product and documentation pages on October 20, 2026. If a vendor has changed a feature, product name, or positioning since then, send a correction and we will update it. Numbered source links in the page body point to the ordered sources below.
Prefactor scores each step of every agent run and can hold or stop an action when a score says it should not happen. Start on the free tier: 5,000 spans a month, no credit card required.
Prefactor helps teams observe, evaluate, and improve their AI agents in production, across every framework and provider.