ReleaseNVIDIANVIDIApublished Oct 27, 2026seen 3h

NVIDIA/NeMo-Relay 0.11.0-rc.3

NVIDIA/NeMo-Relay

Open original ↗

Captured source

source ↗
published Oct 27, 2026seen 3hcaptured 3hhttp 200method plain

0.11.0 RC 3

Repository: NVIDIA/NeMo-Relay

Tag: 0.11.0-rc.3

Published: 2026-10-27T02:38:29Z

Prerelease: yes

Release notes:

What's Changed

⚡ Improvements

  • perf: reuse owned payloads in built-in request codecs by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1493
  • enhancement: reduce ATIF output size by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1520
  • refactor(python): type google adk class swaps by @greysonlalonde in https://github.com/NVIDIA/NeMo-Relay/pull/1505
  • refactor: address Sonar findings by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1530

🐛 Bug Fixes

  • fix(core): bill uncached input tokens in model pricing by @greysonlalonde in https://github.com/NVIDIA/NeMo-Relay/pull/1500
  • fix(codec): preserve input shape for streamed Anthropic tools by @Somay-kousis in https://github.com/NVIDIA/NeMo-Relay/pull/1509
  • fix: accept config level names in binding log helpers by @Somay-kousis in https://github.com/NVIDIA/NeMo-Relay/pull/1504
  • fix: keep real stream metadata after an empty Azure filter chunk by @Somay-kousis in https://github.com/NVIDIA/NeMo-Relay/pull/1508
  • fix: ignore empty or relative XDG base directories by @Somay-kousis in https://github.com/NVIDIA/NeMo-Relay/pull/1513
  • fix: add Python OpenTelemetry plugin file sinks by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1518
  • fix: make worker gRPC message limits configurable by @vantu-fit in https://github.com/NVIDIA/NeMo-Relay/pull/1499
  • fix: reject BigInt in Gemini response decoding by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1521
  • fix: enforce transport policy for automatic OTLP logs and metrics by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1523
  • fix: classify malformed Gemini responses as invalid arguments by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1522
  • fix: accept large Gemini token counts across bindings by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1524
  • fix: honor adaptive_hints inject_header and inject_body_path by @Somay-kousis in https://github.com/NVIDIA/NeMo-Relay/pull/1525
  • fix: warn on unknown keys in [gateway], [upstream] and [agents] by @Somay-kousis in https://github.com/NVIDIA/NeMo-Relay/pull/1511
  • fix(node): isolate throwing codec callbacks by @greysonlalonde in https://github.com/NVIDIA/NeMo-Relay/pull/1529
  • fix: preserve worker streaming errors and routes by @ericevans-nv in https://github.com/NVIDIA/NeMo-Relay/pull/1536
  • fix: keep caller metadata when adaptive hints target a metadata path by @Somay-kousis in https://github.com/NVIDIA/NeMo-Relay/pull/1531
  • fix: unify diagnostics HTTP errors and worker inspection by @bbednarski9 in https://github.com/NVIDIA/NeMo-Relay/pull/1537
  • fix(go): recover callback panics and keep stream finalizer output by @greysonlalonde in https://github.com/NVIDIA/NeMo-Relay/pull/1294

🧰 Maintenance

  • chore: update dependencies flagged by security audits by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1494
  • ci: prune old nightly tags and restrict upstream workflows by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1497
  • chore: upgrade ty to 0.0.91 by @greysonlalonde in https://github.com/NVIDIA/NeMo-Relay/pull/1498
  • chore: handle ty suppressions for optional integration types by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1519
  • chore: address Zstandard security advisories on release/0.10 by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1526
  • chore: explicitly enable uv project management by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1528
  • chore: update Fern to 5.168.0 by @dagardner-nv in https://github.com/NVIDIA/NeMo-Relay/pull/1538

📝 Documentation Updates

  • docs: define pull request etiquette by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1539

Other Changes

  • Forward-merge release/0.10 into main by @willkill07 in https://github.com/NVIDIA/NeMo-Relay/pull/1496
  • Forward-merge release/0.10 into main by @rapids-bot[bot] in https://github.com/NVIDIA/NeMo-Relay/pull/1495
  • Forward-merge release/0.10 into main by @rapids-bot[bot] in https://github.com/NVIDIA/NeMo-Relay/pull/1527

New Contributors

  • @Somay-kousis made their first contribution in https://github.com/NVIDIA/NeMo-Relay/pull/1509
  • @vantu-fit made their first contribution in https://github.com/NVIDIA/NeMo-Relay/pull/1499

Full Changelog: https://github.com/NVIDIA/NeMo-Relay/compare/0.11.0-rc.2...0.11.0-rc.3