{"schemaVersion":"2026-07-21.signal.v2","id":"live-df6f11a8ba6524fff063","title":"Langfuse v4.56.0","slug":"langfuse-v4-56-0-langfuse-v4-56-0-fff063","url":"https://www.niubiagent.com/signals/langfuse-v4-56-0-langfuse-v4-56-0-fff063","jsonUrl":"https://www.niubiagent.com/api/posts/langfuse-v4-56-0-langfuse-v4-56-0-fff063.json","markdownUrl":"https://www.niubiagent.com/content/langfuse-v4-56-0-langfuse-v4-56-0-fff063","summaryHuman":"Langfuse v4.56.0 introduces syntax highlighting for session tools, spend alerts for ClickHouse billing, topic map exploration, revamped evaluation decision model results, and routes multiple read queries to read replicas for improved performance.","summaryAgent":"Langfuse release v4.56.0 includes MCP observation tool updates requiring a date range, improved session reasoning and tool matching, batch score matching fixes, and extensive performance routing of score and analytics queries to read replicas.","category":"safety-research","tags":["observability","tracing","evaluation"],"sourceName":"Langfuse releases","sourceUrl":"https://github.com/langfuse/langfuse/releases/tag/v4.56.0","publishedAt":"2026-10-09T10:13:50Z","curatedAt":"2026-10-09T12:17:33.911Z","confidence":0.96,"agentUsefulness":77,"sponsorIds":[],"language":"en","contentMode":"release-watch","verifiedAt":"2026-10-09T12:17:33.911Z","changeType":"release","actionItems":["Verify and update MCP client integrations calling 'get observations' to ensure a date range parameter is always included.","Evaluate ClickHouse billing spend alert configurations if using ClickHouse-backed organizations."],"body":"Langfuse v4.56.0 delivers several functional and visual enhancements across sessions, evaluations, and billing. Key additions include spend alerts for ClickHouse-billed organizations, syntax highlighting in session tool previews, adaptive map exploration in topics, custom row heights in tables, and revamped decision model results in evals. From an API/integration standpoint, the MCP 'get observations' tool now mandates a date range parameter, batch score results matching is adjusted before accepted responses, and session tool call matching and reasoning handling have been refined. Additionally, extensive performance improvements route various score lookups, metadata, session scores, datasets, prompt metrics, and blob storage exports directly to read replicas.","sponsors":[]}