The Best Open-Source LLM Observability Tools in 2026, Ranked by License Reality
Five open-source LLM observability tools judged on the one thing marketing pages blur - whether "open source" means MIT, Apache-2.0, source-available, or a frozen codebase. One clean winner, one you should not start on.
Published:
“Open source” is the most abused phrase in this category. Every observability vendor claims it. Then you read the license and find a source-available server, or a hobbled core with the real features gated behind a paid cloud, or - in one case here - a repo that is genuinely open but frozen by its new owner. I have self-hosted most of these, and the license page tells you more about what you are signing up for than any feature list.
So that is how I ranked these five. Not on the demo - on what you actually get when you run it yourself.
Four axes, in the order that decides a self-host:
- License reality. MIT and Apache-2.0 are OSI open source. Elastic License 2.0 is source-available. “Frozen” is a category of its own.
- Self-host completeness. Do you keep the real product when you run it yourself, or a stripped core.
- OpenTelemetry support. Native architecture or a bolted-on ingestion path.
- Eval depth. Whether it scores output quality or just shows you what happened.
One clear winner, and one tool I am telling you not to start on.
The short version
| Tool | License | Self-host | OTel | Starting price |
|---|---|---|---|---|
| Langfuse | MIT | Free, near-complete | OTLP backend (HTTP) | Free / $29/mo |
| Opik | Apache-2.0 | Free, full features, no gates | One ingestion path | Free / $19/mo |
| Arize Phoenix | ELv2 server | Free, but source-available | Native | Free (OSS) |
| Laminar | Open source | Free, the whole stack | Native | Free / $30/mo |
| Helicone | Apache-2.0, frozen | Free, but maintenance mode | Partial | Free / $79/mo |
1. Langfuse - the open-source default, and it earns it
Langfuse is the answer most people already give to “what’s the self-hostable observability tool,” and the license backs it up. It is MIT, and in the self-hosted build only three features are enterprise-gated - Organization Creators, the Instance Management API and UI customization. Tracing, evals, prompt management, human annotation and RBAC are all free. That is unusually honest. Most “open-source” tools gate the good parts.
The economics are the other half. At 1M events a month the managed tier runs about $101/mo, versus LangSmith’s roughly $2,514/mo for comparable volume - the widely-cited ~25x gap - and self-hosting removes the per-trace cost entirely. It runs as an OpenTelemetry backend on an OTLP endpoint.
The catch is operational, not commercial. Langfuse v3 moved from “just Postgres” to Postgres plus ClickHouse, Redis and S3-compatible storage - four services to stand up, and the migration to that architecture is where self-hosters get stuck. gRPC OTLP is not supported yet, HTTP only. And it is now a ClickHouse subsidiary after the January 2026 acquisition, which I would file away for a multi-year bet. If you can run the stack, nothing else gives you this much for free.
2. Opik - the same openness, without four services
If Langfuse’s appeal is the openness but the four-service self-host is more than you want to run, Opik is the move. It is Comet’s platform, and it does something no one else here quite matches - the OSS build is Apache-2.0 with the full feature set self-hosted, unlimited spans, members and retention, no gates. That is the most permissive license in the set, cleaner than Phoenix’s Elastic License.
On the cloud, Pro is $19/mo for 100k spans, the cheapest paid tier of the major platforms, with $5 per additional 100k. It is also the fastest-growing project of its peers at roughly 20.8k GitHub stars, and the eval side is real - LLM-as-judge, code metrics, online evaluation, guardrails and an Agent Optimizer.
The gotcha is per-seat pricing at scale. The $19 headline is the small-team configuration, and the recurring complaint is that seat costs climb as the team grows. A few users report UI slowdown on very large projects. OTel is one ingestion path here alongside 60+ framework integrations, not the native architecture. None of that undercuts the core deal.
3. Arize Phoenix - fastest to try, but read the license
Arize Phoenix is the OSS pick that starts fastest - a working trace UI on your laptop in under a minute. It is genuinely OpenTelemetry-native, built on OTel and Arize’s own OpenInference conventions, and its 50+ pre-built eval metrics include what reviewers call the best RAG evaluation in the category.
The gotcha is the license, and it is a real one. Arize markets Phoenix as “fully open source, no feature gates,” but the main server repo is Elastic License 2.0 - source-available, not OSI-approved open source. Only the client and eval subpackages - phoenix-evals, phoenix-client, phoenix-otel - are Apache-2.0. ELv2 forbids offering Phoenix as a hosted service to third parties. For internal use it behaves like open source and the features are not gated. But if your plan is to resell it as a service, the license rules it out. There are also reports of ingest lag before traces appear. Arize the company raised a $70M Series C in February 2025, so it is not going anywhere.
4. Laminar - self-host the whole stack, built for agents
Laminar is the most purpose-built option for AI agents, especially browser agents. It is OpenTelemetry-native, written in Rust for low overhead, and it is the only platform here you can self-host in full - the whole stack is open source, not just a gateway or a hobbled core. Browser Use, one of the most popular open-source browser agents, documents Laminar as its observability integration, and OTel co-creator Ben Sigelman is an angel investor.
The gotcha is the billing, if you use the cloud. You pay on two axes - data by the GB and “Signals,” and Signals are metered by the tokens spent reading your traces, not the tokens your agent spends. That depends on Laminar’s own trace-compression claims, which makes a monthly forecast genuinely hard. Self-hosting removes the usage bill entirely. It is the youngest and smallest here (2024, YC S24, $3M seed), and the exact self-host versus cloud feature delta is not clearly documented, so verify it before you commit. Cloud starts at $30/mo.
5. Helicone - do not start here
I am including Helicone only to tell you not to build on it. It was a clean, Apache-2.0, proxy-based observability tool. Then Mintlify acquired it on 3 March 2026 and put it in maintenance mode - security patches, bug fixes and new-model support only, no roadmap, and Mintlify is actively helping customers migrate off.
For a new buyer that single fact overrides everything else. The Apache-2.0 self-host still works, but self-hosting a frozen codebase whose own creator is pointing people to the exit is a weak bet. The proxy model was also a structural single point of failure - if Helicone is down, your LLM calls fail even when the provider is healthy, and every proxied call adds latency (Helicone cites ~10ms). If you are already on it, use the runway to plan your migration. If you are evaluating fresh, skip to Langfuse or Opik.
So which one?
- You want the best open-source default and can run a four-service self-host - Langfuse. MIT, near-complete, and roughly 25x cheaper than LangSmith at scale.
- You want the same openness without the ops load, or the cheapest managed cloud - Opik at $19/mo, Apache-2.0 with a full-featured self-host.
- You want to try something in the next 60 seconds, or you need the best RAG eval - Arize Phoenix, as long as you are not reselling it and can live with a source-available server.
- You are building agents, especially browser agents, and want to self-host everything - Laminar, self-hosted to sidestep the Signals meter.
- You are on Helicone today - plan your exit. Do not start there fresh.
Every price and date here was read from each vendor’s own materials and links to our full tool reviews. This category ships breaking changes monthly, so we re-verify every 30 days. The one distinction that decides most of these choices - the gap between a permissive license, a source-available one, and a frozen one - is not something a demo will ever show you. Read the license first.
Frequently Asked Questions
What is the best open-source LLM observability tool in 2026?
For most teams it is Langfuse. It is genuinely MIT-licensed, self-hosts free, and only three features are enterprise-gated - tracing, evals, prompt management, human annotation and RBAC are all included. The catch is operational, not commercial - the v3 self-host is four services (Postgres, ClickHouse, Redis and S3-compatible storage). If you want the same openness without running four services, Opik is Apache-2.0 with the full feature set self-hosted and no gates.
Which open-source LLM observability tools are actually OSI open source?
Langfuse (MIT), Opik (Apache-2.0), Laminar (open source, self-host the full stack) and Helicone (Apache-2.0) are OSI-approved. Arize Phoenix is the exception - its server repo is Elastic License 2.0, which is source-available, not OSI open source, and forbids offering Phoenix as a hosted service to third parties. Only Phoenix's client and eval subpackages are Apache-2.0.
Can I self-host these tools with the full feature set?
Opik and Laminar ship the full product self-hosted with no feature gates. Langfuse ships nearly everything - only three minor features are enterprise-gated. Phoenix self-hosts free with no feature gates, but under the Elastic License 2.0 rather than a permissive one. Helicone self-hosts under Apache-2.0 but is in maintenance mode, so you would be running a frozen codebase.
Should I start a new project on Helicone in 2026?
No. Mintlify acquired Helicone on 3 March 2026 and put it in maintenance mode - only security patches, bug fixes and new-model support ship, there is no roadmap, and Mintlify is helping customers migrate off. The Apache-2.0 self-host still works, but building fresh on a frozen tool whose own owner is pointing people to the exit is a dead end. Start on Langfuse or Opik.
Explore More
Related Articles
- The Best LLM Observability Tools in 2026, Ranked and Road-Tested
- The 2026 LLM Observability Consolidation Map - Who Got Bought, Who Stayed Free
- How to Benchmark AI Agents in 2026 - The Tools and the Method
- The Best Free LLM Observability Tools in 2026, Ranked by What "Free" Actually Buys You
- The Best LLM Monitoring Tools in 2026, Ranked for Production Cost and Reliability
Free Newsletter
Get the LLM Evals Newsletter
Platform comparisons, pricing changes and eval technique deep-dives. No spam.
Related Articles
How to Benchmark AI Agents in 2026 - The Tools and the Method
Benchmarking an agent is not benchmarking a model. Public leaderboards tell you about the LLM, not your agent on your task. Here is how to build a real agent benchmark, and the five tools that actually run one - simulation, datasets, trajectory scoring and repeatable eval sets, ranked.
July 26, 2026
best-ofThe Best AI Agent Observability Tools in 2026, Ranked for Multi-Step and Browser Agents
Four platforms for tracing agents that loop, call tools, and click around browsers - judged on agent-native tracing, self-host reality, pricing you can forecast, and pre-release testing. One purpose-built winner, and where each meter bites.
July 26, 2026
best-ofThe Cheapest LLM Observability Tools in 2026, Ranked by Real Cost
The three lowest-cost ways to get production LLM tracing - the cheapest managed cloud, the cheapest self-host, and the free tier that looks great until you read the fine print. Priced at the tiers you will actually hit.
July 26, 2026
Langfuse Review
Opik Review
Arize Phoenix Review
Laminar Review