Langfuse vs LangWatch (2026)

More than 100 integrations with agent frameworks, model providers and gateways send traces into Langfuse, data exports to PostHog, Mixpanel and blob storage, and prompt changes send notices through webhooks and Slack. These move telemetry in and data out, and there are no connectors that let an agent take authenticated actions in outside systems.

SourceLangfuse, langfuse.com homepage and pricingread 2026-09-21

Langy, LangWatch's agent, changes code through pull requests that a GitHub App opens. LangWatch workflows, which can be published as agents or API endpoints, call outside systems through HTTP nodes and code blocks that use encrypted project secrets. The AI Gateway governs which tools and servers a virtual key may reach.

SourceLangWatch, langwatch.ai/docs/llms.txt (Langy pull requests, workflows, secrets, routing policies)read 2026-09-21

For agents that run elsewhere, Langfuse provides tracing, evaluation and prompt management, and prompt composability links prompts together. There are no workflows that sequence, branch or retry an agent's steps, or that mix deterministic nodes with agent steps.

SourceLangfuse, langfuse.com homepage and pricingread 2026-09-21

A LangWatch workflow is a graph of nodes that runs from an Entry point to an End node, mixing LLM prompt nodes with Python code, HTTP and evaluator nodes. Versions are saved and can be restored, and a workflow can be published as an evaluator, as an agent under test, or as an API the customer's code calls. Workflows do not include loops, branching, retries or fallback paths.

SourceLangWatch, langwatch.ai/docs/workflows/overview and llms.txt (building a workflow, workflow as agent); langwatch.ai/docs/workflows/overview.mdread 2026-09-21

Online evaluation runs LLM as a judge and code evaluators on production observations as they arrive, and monitors and alerts fire on metric and evaluator thresholds. Evaluation work therefore starts from incoming telemetry with no person starting each run, and prompt changes also emit webhooks.

SourceLangfuse, langfuse.com homepage and pricingread 2026-09-21

Online evaluation monitors run an evaluator on every matching trace or thread as it arrives, alerts and automations catch regressions and notify teams, automations add every new trace matching a filter to a dataset, and Langy can set up alerts and automations from what it finds across traces. Scoring and actions start on incoming traffic without anyone asking.

SourceLangWatch, langwatch.ai/docs/llms.txt (online evaluation, alerts and automations, datasets from traces, Langy insights)read 2026-09-21

Traces, prompts and evaluation datasets are stored, but there is no retrieval structure over the customer's documents or knowledge that grounds an agent's answers. Langfuse evaluates the grounding of RAG systems built elsewhere instead of providing one.

SourceLangfuse, langfuse.com pricing and llms.txtread 2026-09-21

Built in RAG evaluators score retrieval, and cookbooks on embedding tuning and vector versus hybrid search help with the customer's own pipeline. LangWatch has no document ingestion, index, retrieval layer or knowledge API that grounds an agent in company data.

SourceLangWatch, langwatch.ai/docs/llms.txt (built-in evaluators, cookbooks)read 2026-09-21

Sessions, users and traces are recorded for inspection, and there is no session, conversation or long term memory that an agent reads back as context.

SourceLangfuse, langfuse.com pricing and homepageread 2026-09-21

Traces, threads and datasets are stored as records of the customer's agent, and each Langy conversation runs in its own sandboxed worker. Agents have no session, workflow or long term memory that they read and write across runs.

SourceLangWatch, langwatch.ai/docs/llms.txt (concepts, datasets, how Langy works)read 2026-09-21

Protected deployment labels restrict who can promote a prompt version to production, and CI checks can require an approved baseline before a change ships, so changes to an agent's behavior pass a release gate that people control. There is no approval step, escalation rule or pause before an agent acts at run time.

SourceLangfuse, langfuse.com pricing and llms.txtread 2026-09-21

Guardrails run evaluators inline to block or modify harmful responses in real time, the AI Gateway enforces budgets, rate limits and routing policies, including which tools and servers a key may reach, before a request goes through, and Langy's code changes arrive as pull requests that a person reviews before merge.

SourceLangWatch, langwatch.ai/docs/llms.txt (guardrails, AI Gateway budgets and routing policies, Langy pull requests)read 2026-09-21

Reports for the SOC 2 Type II and ISO 27001 attestations are available from the Pro plan, and there is a region ready for HIPAA and a GDPR DPA. Access control covers RBAC at organization and project level, sign in with Google, Azure AD or GitHub, Enterprise SSO with Okta or Entra ID and SSO enforcement on the Teams add-on, SCIM provisioning, audit logs, data masking on the client side and data retention management.

SourceLangfuse, langfuse.com pricing and homepageread 2026-09-21

Controls include single sign on with directory provisioning and sign in policies, SCIM 2.0 provisioning from Okta, Entra ID and others with group to role mapping, and role based access with custom roles and role bindings at organization, team and project level.

An audit log records every change made through settings, the API and the AI Gateway, with who, where and what changed, and exports to a SIEM in OCSF format. Retention is set per organization, team or project, and secrets are encrypted and write only. LangWatch states ISO 27001 certification and GDPR compliance, with a trust center monitored by Vanta.

SourceLangWatch, langwatch.ai/docs/llms.txt (SSO, SCIM, RBAC, audit log, data retention, secrets, OCSF export), langwatch.ai homepage and pricingread 2026-09-21

Hierarchical traces capture every LLM call, tool invocation and retrieval step of an agent run, so each run's steps and tool calls are visible. They appear as trace trees and agent graphs with session and user tracking, token cost and latency, and can be filtered by user, session, cost, latency or metadata. Audit logs are kept separately on Enterprise, data can be exported in batches or on a schedule to blob storage and to PostHog or Mixpanel, and historical data access runs 30 days on Hobby, 90 days on Core and 3 years on Pro.

SourceLangfuse, langfuse.com homepage and pricingread 2026-09-21

LangWatch traces LLM calls, tool executions and retrieval steps through OpenTelemetry native SDKs and integrations, including coding assistants such as Claude Code and Codex. An audit log records every change through settings, the API and the gateway, governance events export to a SIEM as OCSF records, and analytics export into the customer's dashboards. Default retention is published per plan and can be adjusted per organization, team or project.

SourceLangWatch, langwatch.ai/docs/llms.txt (observability, audit log, OCSF export, data retention) and langwatch.ai/docs/pricingread 2026-09-21

Customers of Langfuse Cloud choose US, EU or JP data regions, and there is a region ready for HIPAA and AWS PrivateLink. The MIT licensed platform self hosts with Docker Compose, Kubernetes via Helm, or Terraform on AWS, GCP and Azure, with paid Enterprise additions for self hosted deployments.

SourceLangfuse, langfuse.com pricing, homepage and llms.txtread 2026-09-21

Customers can run LangWatch as managed multi tenant SaaS in the EU, US, UK or APAC, self hosted with Docker, Kubernetes and Helm or in their own VPC, or hybrid, with the data plane on their infrastructure and the control plane on LangWatch's. The open source repository includes every feature for local use, and an Enterprise license adds longer retention to self hosted deployments.

SourceLangWatch, langwatch.ai homepage and langwatch.ai/docs/llms.txt (self-hosting overview, editions and licensing, hybrid setup)read 2026-09-21

The Langfuse Assistant in the app is the one prebuilt agent, and a Langfuse skill packages best practice workflows for instrumentation, prompt management and API access into a customer's coding agent. Beyond those two, there is no set of prebuilt agents or templates for a customer to choose from.

SourceLangfuse, langfuse.com llms.txt and pricingread 2026-09-21

LangWatch ships Langy, an automated AI engineer that reads traces and evals, writes Scenario tests, optimizes prompts and opens pull requests, and a Skills Directory of installable skills, including skills for product managers and domain experts, that set a coding assistant up to work with LangWatch. There is one packaged agent and a set of setup skills, but no set of ready made agents or workflow templates to choose from.

SourceLangWatch, langwatch.ai/docs/llms.txt (Langy, skills directory) and langwatch.ai homepageread 2026-09-21

Customers configure their own LLM connections, which the Playground uses to test prompts on real production inputs and compare models side by side, and which power the LLM-as-a-judge evaluators the customer sets up. The customer brings its own providers and chooses the models Langfuse's own LLM features run on.

SourceLangfuse, langfuse.com homepage, pricing and llms.txt (LLM Connections)read 2026-09-21

The LangWatch AI Gateway offers one endpoint, compatible with OpenAI and Anthropic, across providers including OpenAI, Anthropic, Azure OpenAI, AWS Bedrock, Vertex AI, Gemini, xAI, Groq, Cerebras and DeepSeek. Provider credentials are held in LangWatch, each application gets a virtual key, and routing policies set which providers a key routes through, the fallback order and model tiers, with budgets and rate limits enforced before the request. Model providers are stored once per organization, team or project for evaluators, workflows and Langy.

SourceLangWatch, langwatch.ai/docs/llms.txt (AI Gateway, routing policies, budgets, providers, model providers)read 2026-09-21

An extensive public API has published rate limits by plan, alongside native Python and TypeScript SDKs, OpenTelemetry ingestion for Java, Go and other languages, a CLI, and a platform MCP server that lets IDE agents manage prompts and query traces. Stable APIs based on IDs manage evaluators, and there is an agent skill.

SourceLangfuse, langfuse.com homepage, pricing and llms.txtread 2026-09-21

LangWatch offers a REST API, including the Workflows and Instant Evals APIs, SDKs in Python, TypeScript, Go and Java, and the langwatch CLI, which drives the platform from the terminal for people and coding assistants. An MCP server gives coding assistants access to traces, tests and evaluations, alongside installable skills, personal and service API keys for development and CI/CD, and CI integration that gates merges on eval results.

SourceLangWatch, langwatch.ai/docs/llms.txt (CLI, MCP server, API keys, API reference) and langwatch.ai/pricingread 2026-09-21

Evaluation of the customer's application runs online and offline. It covers datasets, experiments from the SDK or UI compared side by side against baselines, LLM as a judge and code evaluators managed by API, custom scores, user feedback and human annotation queues, alerts on evaluator results, and CI checks that block regressions against thresholds and approved baselines. The customer's agent is tested with datasets before production, and output quality is scored over time.

SourceLangfuse, langfuse.com llms.txt, homepage and pricingread 2026-09-21

Experiments run over datasets with built in, saved and workflow built evaluators. LangWatch also tests agents with multi turn scenarios in which a simulated user acts against the agent and a judge scores the conversation. It groups scenarios into test suites, compares agents side by side on pass rate, cost and latency, gates merges on eval results in CI, scores production traffic with online monitors, and optimizes prompts and tool contracts with the scenario suite as the quality gate.

SourceLangWatch, langwatch.ai/docs/llms.txt (agent testing, evaluations, improve your agent) and langwatch.ai/pricingread 2026-09-21

Langfuse observes and evaluates agents, and there is no browser, desktop or computer control by an agent.

SourceLangfuse, langfuse.com homepageread 2026-09-21

The product covers tracing, agent testing, evaluation, the AI Gateway, governance, Langy and self hosting, and no agent controls a browser, desktop or computer. Langy works in a sandbox through the CLI and code.

SourceLangWatch, langwatch.ai/docs/llms.txtread 2026-09-21