← Back to Blog
Alternatives guidePublished July 13, 2026· Last verified July 15, 2026

Langfuse Alternatives (2026): 6 LLM Observability Options

Langfuse is the most adopted open-source LLM observability platform, and since January 2026 it belongs to ClickHouse. For most teams it remains the right default. This page is for the teams where it is not — and it is honest that the last entry on the list solves a different problem.

Langfuse became the open-source default for LLM tracing on merit: an MIT core with ~31k GitHub stars, a self-hosted build that runs the exact same codebase as Langfuse Cloud, and steady 2026 shipping — first-class Experiments, CI/CD quality gates, Monitors & Alerts. In January 2026 ClickHouse acquired the company, announcing that the roadmap, MIT licensing, self-hosting and Langfuse Cloud continue unchanged (acquisition announcement).

So why an alternatives page? Because ownership changes make prudent buyers re-underwrite, because some of Langfuse's governance features are Enterprise-licensed even when self-hosted, and because Langfuse is deliberately observe-only — it watches the runtime, it does not police it. Six alternatives follow, ranked, each scored on the same ten axes and cited to primary sources dated July 15, 2026.

Why teams look beyond Langfuse

Start with what Langfuse gets right, because most of this list does not beat it on its home turf. The MIT core is genuinely unlimited when self-hosted, the networking docs state it runs without internet access, and the eval stack — LLM-as-a-judge, code evaluators, Experiments with CI/CD gates, annotation queues — is deep for an open-source tool. EU-default hosting and ISO 27001/SOC 2 Type II audits make the procurement conversation short.

The first reason teams look elsewhere is the ownership change. ClickHouse committed publicly to continuity in January 2026, and nothing in the changelog since suggests otherwise — but a database company now owns the roadmap, and buyers who were betting on Langfuse's independence are entitled to re-check that bet.

The second is the open-core boundary. Audit logs, data-retention policies, project-level RBAC, SCIM and server-side masking — precisely the features a governance-minded self-hoster wants — require an Enterprise license key, and the EE license telemetry cannot be disabled (telemetry docs). The core is MIT; the governance layer is not.

The third is architectural, and Langfuse is admirably honest about it: it is observe-only. Its own docs delegate runtime blocking — prompt injection, PII, toxicity — to third-party libraries and position Langfuse as ex-post evaluation (security & guardrails doc). Teams that need enforcement in the request path — guardrails, budgets that block rather than alert, agent containment — need a different class of product in front of the model, whatever they keep for tracing.

How we chose the alternatives

  • Tracing and cost-tracking depth on production LLM and agent traffic.
  • Evaluation tooling: judges, datasets, experiments, human annotation, CI integration.
  • Open-source terms and self-hosting reality: license, feature gates, air-gap feasibility.
  • Runtime posture: whether the tool can enforce anything, or only observe.
  • Vendor trajectory in 2026: ownership, funding and shipping velocity, cited and dated.
  • Honest category fit relative to Langfuse's observe-only design.

The alternatives at a glance

ProductBest forDeploymentOpen sourcePricing model
LangSmith (LangChain)AI engineering teams — especially LangChain/LangGraph shops — that want tracing, evals and agent deployment in one platform.SaaS (US, EU, APAC regions), hybrid, or self-hosted Kubernetes (Enterprise add-on)Platform proprietary; SDKs and the LangChain/LangGraph frameworks are MITFree Developer tier; Plus at $39/seat/mo plus per-trace usage; Enterprise by quote (gates self-hosting)
Arize AIAI engineering and platform teams that want enterprise-grade eval rigor with an open-source on-ramp (Phoenix) and a self-hosted option.AX Free/Pro are SaaS; AX Enterprise as SaaS or self-hosted (Kubernetes-first, major clouds and private VPC); Phoenix OSS runs anywherePhoenix under Elastic License 2.0; Arize AX platform proprietaryAX Free tier; AX Pro at $50/mo; AX Enterprise by quote (SaaS or self-hosted); Phoenix OSS free
Opik (Comet)Cost-conscious engineering teams and OSS-first organizations that want self-hosted LLM tracing and evals without per-seat pricing.Self-hosted OSS (Docker Compose or Kubernetes) with no feature restrictions, or Comet-managed cloud; Enterprise offers custom hostingApache-2.0OSS self-host free with unlimited spans; managed cloud Free tier, Pro at $39/mo with unlimited team members, Enterprise by quote
HeliconeExisting Helicone users who need continuity; the maintenance-mode status makes it hard to recommend for new adoption in 2026.SaaS; full platform self-hostable (Docker Compose, Helm); no air-gapped option documentedApache-2.0 (main platform repo, ~5.9k stars)Free Hobby tier (10k requests/month); Pro from $79/mo; Team $799/mo adds SOC 2 and HIPAA; gateway usage billed at provider cost (zero markup).
Weights & Biases WeaveML and AI engineering teams — especially existing W&B experiment-tracking users and CoreWeave customers — adding LLM/agent tracing and evals to an established ML workflow.Multi-tenant SaaS (Free/Pro); Enterprise offers Dedicated Cloud or customer-managed self-hosted deploymentWeave SDK Apache-2.0; W&B platform proprietaryFree tier; Pro with usage-based Weave data ingestion; Enterprise custom (dedicated or customer-managed)
KosmoyRegulated enterprises that need governance enforced in the runtime path, in their own infrastructure.Self-hosted — single-tenant, your own Kubernetes (air-gap capable)ProprietaryEnterprise subscription; no self-service tier.

Last verified July 15, 2026 against each vendor's public documentation.

Capability shape vs Langfuse

Each panel shows one alternative across the same ten capability axes (0–10); the dashed outline is Langfuse for reference. The further a shape reaches on a spoke, the stronger that capability.

LangSmith (LangChain)
INVSECOBSGWGRDCTNCMPEVLBLDSOV
Arize AI
INVSECOBSGWGRDCTNCMPEVLBLDSOV
Opik (Comet)
INVSECOBSGWGRDCTNCMPEVLBLDSOV
Helicone
INVSECOBSGWGRDCTNCMPEVLBLDSOV
Weights & Biases Weave
INVSECOBSGWGRDCTNCMPEVLBLDSOV
Kosmoy
INVSECOBSGWGRDCTNCMPEVLBLDSOV
dashed = LangfuseINV AI Inventory & Discovery · SEC Security & Shadow AI · OBS Observability & FinOps · GW Gateway & Policy Control · GRD Guardrails & Runtime Safety · CTN Agent Containment · CMP Compliance & Audit · EVL Testing, Evals & Red-teaming · BLD Agent Building · SOV Deployment Sovereignty

The alternatives, one by one

01

LangSmith (LangChain)

LLM observability, evals & agent engineering platform

LangSmith is LangChain's commercial platform for agent engineering — tracing, evaluation, prompt management, agent deployment, sandboxes and a no-code agent builder, plus an LLM gateway in private beta — layered on the MIT-licensed LangChain and LangGraph frameworks.

The most complete platform alternative: everything Langfuse does plus agent deployment, sandboxes, no-code agents and Engine automation — if you accept a proprietary platform and an Enterprise-gated self-host.

Where it beats Langfuse

  • Platform breadth no one else matches in 2026: LangSmith Deployment, Fleet, Sandboxes GA, Engine public beta and an LLM gateway in private beta.
  • Eval automation a step beyond Langfuse: Engine clusters recurring production issues, proposes fixes and auto-generates evaluators and datasets.
  • The LangChain/LangGraph ecosystem (~141.8k and ~37.3k GitHub stars) surrounds the platform with integrations and hiring-pool familiarity.

Where it falls short

  • The platform is proprietary; self-hosting is an Enterprise add-on with a license key and, outside air-gapped licenses, mandatory egress to beacon.langchain.com — Langfuse self-hosts under MIT with no gate.
  • Seat-plus-usage pricing with quote-gated Enterprise, against Langfuse's published tiers and free self-host.
  • For EU-resident buyers, Langfuse's Berlin heritage and EU-default region remain the simpler story.
Deployment: SaaS (US, EU, APAC regions), hybrid, or self-hosted Kubernetes (Enterprise add-on)Open source: Platform proprietary; SDKs and the LangChain/LangGraph frameworks are MITPricing: Free Developer tier; Plus at $39/seat/mo plus per-trace usage; Enterprise by quote (gates self-hosting)
02

Arize AI

AI observability & evaluation platform (Arize AX + Phoenix OSS)

Arize AI pairs the enterprise Arize AX platform (agent observability, evaluation and runtime guards, SaaS or self-hosted) with Phoenix, one of the most active open-source AI observability projects.

The evaluation heavyweight: versioned evaluators, production trace review and runtime guards across the enterprise AX platform and the open-source Phoenix project.

Where it beats Langfuse

  • Evaluator Hub versions LLM-as-a-judge evaluators at commit level — evaluation as a governed artifact, beyond Langfuse's evaluator library.
  • Runtime Guards can block, substitute or regenerate responses; Langfuse deliberately leaves blocking to third parties.
  • Voice-agent observability with audio session replay, undocumented in Langfuse as of July 15, 2026.

Where it falls short

  • Phoenix is Elastic License 2.0, not OSI-approved — a real difference for teams that chose Langfuse for its MIT core.
  • The full AX platform is proprietary with quote-based Enterprise; air-gapped deployment is not documented.
  • No equivalent of Langfuse's transparent, published cloud pricing ladder.
Deployment: AX Free/Pro are SaaS; AX Enterprise as SaaS or self-hosted (Kubernetes-first, major clouds and private VPC); Phoenix OSS runs anywhereOpen source: Phoenix under Elastic License 2.0; Arize AX platform proprietaryPricing: AX Free tier; AX Pro at $50/mo; AX Enterprise by quote (SaaS or self-hosted); Phoenix OSS free
03

Opik (Comet)

Open-source LLM evaluation & observability platform

Opik is Comet's Apache-2.0 open-source platform for tracing, evaluating, monitoring and optimizing LLM applications and agents — self-hosted with no feature restrictions, or as Comet-managed cloud.

The closest like-for-like open-source swap: Apache-2.0, ~20.6k stars, unlimited self-hosted spans with no feature restrictions, plus guardrails in beta.

Where it beats Langfuse

  • License terms: Apache-2.0 with no EE carve-out documented — the governance features Langfuse gates behind an Enterprise key have no equivalent paywall in Opik's OSS build.
  • Guardrails (beta) ship in the toolkit — Presidio-based PII filtering and topic moderation — where Langfuse offers none.
  • Managed-cloud pricing undercuts Langfuse: Pro at $39/mo with unlimited team members, versus Langfuse's $199/mo Pro tier.

Where it falls short

  • Smaller certification posture: no SOC 2 Type II / ISO 27001 story equivalent to Langfuse's published audits.
  • Guardrails remain beta, with no prompt-injection blocking documented as of July 15, 2026.
  • Comet's ~$70M funding (last raise 2021) is thinner backing than ClickHouse's ownership of Langfuse.
Deployment: Self-hosted OSS (Docker Compose or Kubernetes) with no feature restrictions, or Comet-managed cloud; Enterprise offers custom hostingOpen source: Apache-2.0Pricing: OSS self-host free with unlimited spans; managed cloud Free tier, Pro at $39/mo with unlimited team members, Enterprise by quote
04

Helicone

LLM observability platform with an AI gateway

Helicone is an open-source, developer-first LLM observability platform (traces, costs, prompts, experiments) with a lightweight Rust AI gateway and zero-markup cloud gateway — in maintenance mode since Mintlify acquired the company in March 2026, with feature development ended.

The proxy-based logging pioneer — but in maintenance mode since the March 2026 Mintlify acquisition: fine for existing users, a risk for new adoption.

Where it beats Langfuse

  • Integration by base-URL change: the proxy captures traffic without SDK instrumentation, faster to adopt than Langfuse's SDK/OTel path.
  • A gateway line Langfuse never built: an open-source Rust gateway plus a cloud gateway with zero-markup passthrough billing.
  • Sits in the data path, so rate and cost limits can act on traffic rather than only report it.

Where it falls short

  • Feature development ended with the Mintlify acquisition (March 3, 2026); only security patches, bug fixes and new-model support continue.
  • Evals are thin next to Langfuse's judges, datasets, experiments and CI/CD gates.
  • No guardrails, containment or compliance tooling.
Deployment: SaaS; full platform self-hostable (Docker Compose, Helm); no air-gapped option documentedOpen source: Apache-2.0 (main platform repo, ~5.9k stars)Pricing: Free Hobby tier (10k requests/month); Pro from $79/mo; Team $799/mo adds SOC 2 and HIPAA; gateway usage billed at provider cost (zero markup).
05

Weights & Biases Weave

LLM/agent tracing & evaluation within the W&B platform

W&B Weave is Weights & Biases' toolkit for tracing, evaluating and monitoring LLM and agent applications — agent-first trace semantics on top of the W&B ML platform, owned by CoreWeave since May 2025.

The right answer inside a Weights & Biases estate: agent-first tracing and mature evals joined to ML experiment tracking, under CoreWeave ownership since May 2025.

Where it beats Langfuse

  • Agent-first trace semantics — sessions, turns, steps, tools, sub-agents — where Langfuse's model is more generic.
  • One platform from model training to LLM app traces, a consolidation Langfuse cannot offer.
  • Pre-built inline safety scorers (toxicity, PII, prompt injection) available today.

Where it falls short

  • The platform is proprietary; only the Weave SDK is Apache-2.0, and self-managed feature parity is not documented as of July 15, 2026.
  • Guardrail enforcement is wired per application in code — no central policy point.
  • CoreWeave ownership orients the roadmap toward its AI cloud; Langfuse's OSS neutrality is part of why teams picked it.
Deployment: Multi-tenant SaaS (Free/Pro); Enterprise offers Dedicated Cloud or customer-managed self-hosted deploymentOpen source: Weave SDK Apache-2.0; W&B platform proprietaryPricing: Free tier; Pro with usage-based Weave data ingestion; Enterprise custom (dedicated or customer-managed)
06

Kosmoy

AI management platform

A self-hosted control plane for enterprise AI: one inventory, one policy gateway, one audit trail and a containment sandbox for every model, agent and MCP server a company runs.

A different category — choose Kosmoy when the problem is governance and audit, not evals. Ranked last on purpose: it does not match Langfuse's tracing and evaluation depth, and it is not trying to.

Where it beats Langfuse

  • Enforcement where Langfuse observes: one OpenAI-compatible gateway applies PII, toxicity, prompt-injection and policy guardrails, RBAC and blocking budgets to every LLM, MCP and A2A call.
  • Org-wide AI inventory with risk tiers and owners, including a master agent registry fed by Azure AI Foundry, Bedrock, Vertex, Salesforce and ServiceNow connectors — Langfuse has no inventory at all.
  • EU AI Act, ISO/IEC 42001 (aligned) and NIST AI RMF evidence bundles, plus kernel-enforced agent sandboxing with a kill switch (Action Capsule) — none documented by Langfuse as of July 15, 2026.
  • Sovereignty without an open-core boundary: single-tenant in your Kubernetes, air-gap capable, no vendor control plane and no mandatory telemetry.

Where it falls short

  • Evals score 4 on our own radar — no LLM-as-judge, datasets or experiments; Langfuse is categorically better at evaluation, and Kosmoy customers keep a tracing/evals tool alongside it.
  • Operational monitoring (cost, usage, quality, budget alerts) rather than Langfuse-grade development tracing.
  • Proprietary and enterprise-only — no free tier, no self-service, no open-source core.
Deployment: Self-hosted — single-tenant, your own Kubernetes (air-gap capable)Open source: ProprietaryPricing: Enterprise subscription; no self-service tier.

Decision guide

You want more platform — agent deployment, sandboxes, automationLangSmith (Enterprise if you need to self-host it).
You want to stay open source with the fewest license stringsOpik (Apache-2.0, no feature gates self-hosted).
You need versioned evaluators and runtime guards from one vendorArize AX, starting from Phoenix if you want the OSS path.
Your organization already runs W&B for ML workW&B Weave.
You are on Helicone todayNo urgent exit, but plan one — feature development ended with the March 2026 Mintlify acquisition.
An auditor, regulator or agent incident triggered this searchKosmoy for the control plane — and keep Langfuse for tracing; they are complements, not substitutes.

Questions buyers ask

Is LangSmith better than Langfuse?

On platform breadth, yes: LangSmith adds agent deployment, sandboxes, a no-code builder and eval automation that Langfuse does not attempt, and its evaluation suite is the category's most complete. Langfuse wins on openness and self-hosting: an MIT core with no Enterprise gate, the same codebase as its cloud, documented offline operation and published pricing. Ecosystem-heavy teams tend to LangSmith; OSS-first and EU-resident teams tend to Langfuse.

What happened to Langfuse after the ClickHouse acquisition?

ClickHouse acquired Langfuse in January 2026, announced alongside ClickHouse's $400M Series D. Langfuse states the roadmap, MIT licensing, self-hosting and Langfuse Cloud continue unchanged, and its 2026 changelog since the deal — a CLI, rebuilt Experiments, CI/CD gates, Monitors & Alerts, a Japan region — supports that. The honest caveat is simply that long-term direction is now set inside ClickHouse.

Can Langfuse block prompts or enforce guardrails at runtime?

No, by design. Langfuse's own documentation delegates runtime blocking — prompt injection, PII, toxicity — to third-party libraries and positions Langfuse as ex-post monitoring and evaluation of those controls. If you need enforcement, put a policy gateway in the request path and keep Langfuse for tracing; Kosmoy's [gateway](/platform/ai-gateway/) plays that enforcement role in regulated stacks.

What does Langfuse cost?

Self-hosting the MIT core is free with all core features unlimited. Langfuse Cloud runs from a free Hobby tier through Core at $29/mo, Pro at $199/mo and Enterprise at $2,499/mo, metered on usage units; self-hosted Enterprise (the EE license for audit logs, retention, project RBAC and SCIM) is quoted — see [Langfuse's pricing page](https://langfuse.com/pricing) for current terms.

Is Kosmoy a Langfuse alternative?

Only in the sense that both watch LLM traffic. Langfuse is a development tool: traces, evals, prompts. Kosmoy is a governance platform: inventory, gateway-enforced policy, agent containment and [compliance evidence](/platform/ai-compliance/), self-hosted single-tenant in your Kubernetes. If your problem is debugging quality, choose Langfuse. If your problem is proving control to an auditor — with Article 50 EU AI Act transparency obligations applying from August 2, 2026 — choose Kosmoy, and expect to run both.

Sources

Every factual claim about another vendor on this page traces to that vendor's own published material or a named third-party source below.

  1. Langfuse joins ClickHouse (acquisition announcement) — accessed July 15, 2026
  2. Langfuse enterprise license key (EE feature list) — accessed July 15, 2026
  3. Langfuse security & guardrails doc (observe-only positioning) — accessed July 15, 2026
  4. Mintlify acquires Helicone (March 3, 2026) — accessed July 15, 2026
  5. Kosmoy AI Gateway — accessed July 15, 2026
  6. Langfuse GitHub repository (MIT core, stars) — accessed July 15, 2026
  7. Self-hosting overview (same codebase as Cloud) — accessed July 15, 2026
  8. Networking (no internet access required) — accessed July 15, 2026
  9. Telemetry (EE license telemetry cannot be disabled) — accessed July 15, 2026
  10. Langfuse pricing — accessed July 15, 2026
  11. Changelog (Experiments, CI/CD gates, Monitors & Alerts) — accessed July 15, 2026
  12. LangSmith self-hosted overview (docs) — accessed July 15, 2026
  13. LangSmith self-hosted egress & air-gapped licensing (docs) — accessed July 15, 2026
  14. LangSmith LLM Gateway (docs, private beta) — accessed July 15, 2026
  15. LangSmith Sandboxes (docs) — accessed July 15, 2026
  16. LangSmith Fleet overview (docs) — accessed July 15, 2026
  17. LangSmith Deployment overview (docs) — accessed July 15, 2026
  18. Interrupt 2026 launches (LangChain blog) — accessed July 15, 2026
  19. LangSmith pricing — accessed July 15, 2026
  20. Fortune — LangChain raises $125M at $1.25B valuation — accessed July 15, 2026
  21. Phoenix GitHub repository — accessed July 15, 2026
  22. Arize AX self-hosting docs — accessed July 15, 2026
  23. Arize AX guardrails docs — accessed July 15, 2026
  24. Observe 2026 / Arize AX launches (blog) — accessed July 15, 2026
  25. Arize AX release notes — accessed July 15, 2026
  26. Arize pricing — accessed July 15, 2026
  27. Series C press release ($70M) — accessed July 15, 2026
  28. Opik GitHub repository (Apache-2.0) — accessed July 15, 2026
  29. Opik FAQ (deployment, no feature restrictions) — accessed July 15, 2026
  30. Comet pricing — accessed July 15, 2026
  31. Opik Guardrails beta announcement — accessed July 15, 2026
  32. Comet company profile (funding) — accessed July 15, 2026
  33. Helicone main GitHub repo — accessed July 15, 2026
  34. Helicone AI Gateway repo (Rust) — accessed July 15, 2026
  35. Helicone pricing — accessed July 15, 2026
  36. Helicone — joining Mintlify (March 2026) — accessed July 15, 2026
  37. Mintlify acquires Helicone (March 2026) — accessed July 15, 2026
  38. Helicone cloud gateway / passthrough billing launch — accessed July 15, 2026
  39. Weave GitHub repository (Apache-2.0 SDK) — accessed July 15, 2026
  40. CoreWeave completes acquisition of Weights & Biases — accessed July 15, 2026
  41. W&B pricing — accessed July 15, 2026
  42. Weave guardrails and monitors docs — accessed July 15, 2026
  43. CoreWeave + W&B new products press release — accessed July 15, 2026
  44. Weave production-agent observability announcement — accessed July 15, 2026
  45. Kosmoy Platform — accessed July 15, 2026
  46. Kosmoy Action Capsule — accessed July 15, 2026
  47. Kosmoy AI Compliance — accessed July 15, 2026

Need governance, not just a swap?

Kosmoy puts an inventory, a policy gateway and a containment sandbox around every AI your teams run — in your own Kubernetes.

Contact us

Or email sales@kosmoy.com.