Langfuse v4.51.0
Langfuse v4.51.0 introduces MCP decision-model evaluators, direct annotations from the experiment comparison grid, gateway request correlation IDs, and OpenAI Chat Completions relaying with usage telemetry in the AI Gateway.
Why this signal matters
Langfuse version 4.51.0 brings several features and extensive bug fixes across its platform. Key additions include support for decision-model evaluators in MCP, direct annotation capabilities within the experiment comparison grid, persistent resizable charts, and actor information logging for audited actions and deletions. The AI Gateway now supports correlating requests with a gateway request ID and relays OpenAI Chat Completions with usage telemetry. Numerous fixes were applied to standardize API responses (returning 404s for missing datasets, API keys, score configs, and widgets), fix OpenTelemetry batch preservation with valueless attributes, and adjust token count filter comparisons to Float64.
Actionable summary
Langfuse v4.51.0 adds support for MCP decision-model evaluators, correlates AI gateway requests via gateway request IDs, relays OpenAI Chat Completions with telemetry, and improves API error handling (standardizing 404/precondition responses across missing entities).
- Agent usefulness
- 77/100
- Confidence
- 96%
- Canonical data
- JSON + Markdown
What builders should check
- Upgrade Langfuse instances to v4.51.0.
- Verify integrations relying on error handling around deleted or missing resources (e.g., datasets, API keys, score configs) to account for newly standardized 404/not-found responses.
- Adopt the gateway request ID for correlated tracing in AI Gateway pipelines if applicable.