Skip to content
Neutree Documentation

What's in this release

  • Adds the vLLM 0.31.0 engine, incorporating the latest upstream model compatibility and performance improvements.
  • API keys support setting token quotas per model, and the model selector displays model source tags.
  • External endpoints support configuring multiple upstream targets for exposed models and routing requests by using either an active-standby or weighted routing strategy.
  • On the endpoint details page, you can navigate to the corresponding details from the model registry name and model name.
  • Tooltips in the model usage trend chart might have overflowed when many API keys were active. The issue has been resolved in this release. This release also optimizes single-day usage viewing and data pagination.
  • The page displayed the raw error, an empty chart, filter conditions, and an empty list when the user did not have permission to view access logs. The issue has been resolved in this release.