Langfuse roadmap

Langfuse is open source, and we want to be transparent about where the project is going. This page summarizes the product areas we are currently investing in.

Current focus areas

Our goal is to make Langfuse the best open platform for understanding, evaluating, and improving AI agents (see Why Langfuse?).

Langfuse Gateway

Bring model access, observability, and governance into one Langfuse-native control plane.

• Provide a thin, bring-your-own-key gateway with virtual keys and access controls across Langfuse Cloud and self-hosted deployments.
• Connect requests directly to Langfuse tracing and cost tracking, with clear visibility into model usage and spend.

Proactive issue detection

Help teams find the problems that matter without manually reviewing every trace.

• Build a consistent representation of agent traces, including long-running and multi-span interactions.
• Group interactions into recurring topics, behaviors, and failure modes, and surface the most important issues with useful summaries and examples.
• Make it easier to turn production issues into datasets, evaluations, experiments, and improvement workflows.

Better evals and experiments

Extend evaluation and experimentation workflows for increasingly complex agent systems.

• Improve the evaluator foundation and use AI to generate code-based evaluators for specific use cases.
• Evaluate complete agent trajectories and other multi-span interactions.
• Make it easier to compare prompts, models, and runtime changes, including experiments on production traffic.

Scale and enterprise controls

Expand the capabilities teams use to operate Langfuse at increasing scale in the cloud or on their own infrastructure.

• Unify query behavior and improve ingestion reliability, performance, and scale for large agent workloads.
• Make model pricing and regional matching more accurate, and clearly flag missing price definitions.
• Improve API key management, role-based access controls, self-hosting, security, and compliance.

A clearer agent-engineering workflow

Make Langfuse easier to adopt and more coherent across the full improvement loop.

• Create clearer onboarding, home, agent, score, and experiment views.
• Make Langfuse capabilities available through MCP so external agents can use them in their workflows.
• Connect tracing, datasets, evaluations, and experiments with fewer manual steps.

Recently released

The 10 most recent changelog updates:

• Pulse: find the outliers in your traces(Jul 28, 2026)
• Secure remote experiment triggers(Jul 28, 2026)
• Any table is a chart(Jul 24, 2026)
• Keep large observation content inspectable(Jul 24, 2026)
• Track and alert on boolean scores(Jul 21, 2026)
• Manage dashboards via API, CLI, and MCP(Jul 17, 2026)
• Start with root observations(Jul 15, 2026)
• Graph View: Aggregated and Expanded modes(Jul 13, 2026)
• Filter by boolean scores(Jul 10, 2026)
• Evaluate tool calls(Jul 10, 2026)

Shape the roadmap

Your feedback helps us decide what to prioritize:

• Suggest a feature or improvement, or vote on ideas from other users.
• Report a bug.
• Discuss your use case with the community on Discord.
• Join a community hour to talk with the Langfuse team.

---

---