# Langfuse v4.74.0 adds per-organization rate limits and project transfers

published: 2026-11-03T14:05:42.256Z
topic: infrastructure
source: https://github.com/langfuse/langfuse/releases/tag/v4.74.0
url: https://aiagentnews.io/stories/langfuse-v4-74-0-adds-per-organization-rate-limits-and-project-transfers

Langfuse has released v4.74.0 on GitHub, adding per-organization rate limits to the AI Gateway and a way to move projects between organizations. The release also hides shut-down and retired model names in the AI Gateway and includes several fixes to the web UI and sessions.

Langfuse published v4.74.0 on 3 November. The release adds per-organization request and byte rate limits at /resolve in the AI Gateway, and it combines pending Topics schema migrations into one change.

The admin API can now transfer projects between organizations. Support access reads are always recorded in audit logs and marked in ClickHouse, and the release includes tests for changes to the cross-organization access-log setting.

Fixes in this release hide shut-down OpenAI and retired Anthropic models in the AI Gateway and show the AI Gateway banner once the v4 migration banner is dismissed. The internal session customization dialog also received a fix. Build and tooling changes include moving to oxlint and returning to Red Hat's nodejs:24 in the Docker build.

## Why it matters if you build agents

Teams routing agent model calls through the AI Gateway can now set request and byte limits for each organization at /resolve. Code that depends on OpenAI or Anthropic model names should be checked, since shut-down and retired models are no longer offered in the gateway. Teams that reorganize work across accounts can move projects through the admin API rather than recreating them.