Langfuse vs Helicone in 2026 - Why One of These Is a Dead End

If you are comparing Langfuse and Helicone, you are looking at two open-source LLM observability tools that you can self-host for free. On paper that makes it a close call. In mid-2026 it is not.

Mintlify acquired Helicone on 3 March 2026 and put it in maintenance mode. Only security patches, bug fixes and new-model support ship now. No roadmap, no new features, and Mintlify has said it is helping existing customers migrate off. Langfuse, meanwhile, is one of the most active projects in the category. That single fact does most of the work in this comparison, but not all of it - the proxy model Helicone pioneered is genuinely different, and worth understanding before you decide. I have added Opik as the third option, because it is where most people leaving Helicone should actually land.

The short version

Tool	Best for	Pricing model	Self-host / OSS
Langfuse	The all-round open-source default	Free / $29/mo, billable units	Free, MIT, near-complete
Helicone	Nobody new - maintenance mode	Free / $79/mo, usage-based	Free, Apache-2.0, but frozen
Opik	Cheapest managed cloud, permissive OSS	Free / $19/mo, spans	Free, Apache-2.0, full features

Langfuse: the healthy open-source default

Langfuse is the tool most people mean when they ask “what is the self-hostable alternative.” It does tracing, evals, prompt management and human annotation, and it does them framework-agnostic. The reason it wins is that self-hosting gets you the real product - only three features are enterprise-gated, so tracing, evals, prompt management and RBAC are all free under MIT. Most “open-source” tools give you a hobbled core. Langfuse does not.

The economics are the other half. At 1M events a month it runs about $101/mo managed, and self-hosted the per-trace cost disappears. The Core cloud tier is $29/mo for 100k billable units with 90-day retention, and the free Hobby tier gives you 50k units on a hard cap.

The catch is operational, not commercial. Langfuse v3 moved to Postgres plus ClickHouse plus Redis plus S3-compatible storage - four services to run - and the migration to that architecture is where self-hosters get stuck. There are GitHub threads reporting memory errors and timeouts. One more thing to file away: Langfuse was acquired by ClickHouse on 16 January 2026. The MIT license and self-host survived, but you are now betting a multi-year stack on a database company’s subsidiary. Details are on the Langfuse page.

Helicone: clean product, frozen future

I will be fair to Helicone, because the product was good. It is the lightest-weight of the major observability tools - a logging proxy first. The defining feature is the proxy model: you change your API base URL to route through it, and it logs everything with zero SDK and zero code changes, in any language, against any provider. That is genuinely faster to wire up than instrumenting your code. There is also an async SDK logging mode for Python and JS when you do not want the proxy in your critical path. It is Apache-2.0 and self-hosts free via Docker or Helm. By the time Mintlify bought it, it had processed 14.2 trillion tokens across roughly 16,000 organizations.

Now the two gotchas. The first is structural: the proxy sits in the hot path of every LLM call, which makes it a single point of failure. If Helicone is down, your calls fail even when OpenAI or Anthropic are healthy, and every proxied call adds latency - Helicone cites around 10ms, which compounds for real-time apps.

The second overrides everything. Helicone is in maintenance mode, and its own new owner is pointing customers to the exit. Self-hosting an open-source tool is normally a bet that upstream keeps improving. Here, upstream is frozen. That turns “free and open” into “a codebase you maintain that will not get new features.” For a new buyer, that is disqualifying no matter how tidy the tool is.

Opik: the option most Helicone leavers actually want

If the appeal of Helicone was “open source, cheap, low-friction,” Opik is the modern version of that pitch. It is Comet’s open-source observability and eval platform, Apache-2.0 with the full feature set self-hosted - unlimited spans, members and retention, no gates. On the cloud, Pro is $19/mo for 100k spans, the cheapest paid cloud 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 ships 60+ framework integrations plus an OTLP ingestion path.

The honest gotcha is per-seat pricing at scale - the $19 headline is the small-team configuration, and seat costs climb as the team grows. A few users report UI slowdown on very large projects. Neither undercuts the core value, but model the seat cost if you are a big team.

Langfuse vs Helicone: which should you pick?

• You are evaluating fresh in 2026 - Langfuse. It is actively developed, MIT-licensed, and the self-host is genuinely complete. Helicone’s maintenance-mode status takes it off the table for new builds.
• You want the least-friction setup and zero SDK - Helicone’s proxy did this best, but the product is frozen. Get the same low-friction feel from Opik or Langfuse’s OTLP backend instead, both of which are still shipping.
• You want the cheapest managed cloud - Opik at $19/mo, ahead of Langfuse Core at $29/mo.
• You can run a four-service self-host stack - Langfuse, where self-hosting removes the per-trace cost entirely.
• You are already on Helicone today - use the runway to plan your migration. Do not build anything new on it.

This is the rare comparison with a clear answer. Helicone was a clean tool in a different timeline. But a frozen product whose own owner is helping people leave is not something to start on in 2026. Every price and date here was read from each vendor’s own materials, verified on 26 July 2026, and this category ships breaking changes monthly, so we re-verify every 30 days. Between these two, pick Langfuse - and if managed and cheap is the goal, look hard at Opik.