Starstruk Logo MarkStarstruk

v1.2.1

alibaba/MNNv1.2.1Apr 7, 2026by PawelPeczek-Roboflow

AI Summary

Adds significant workflow capabilities including Vision Events blocks and CUDA memory profiling. Introduces SAM3 support with class_mapping and model introspection features for cached models.

Key Highlights

  • Added Workflow Profiling documentation
  • Added Roboflow Vision Events workflow block
  • Added CUDA memory leak profiling script
  • Added Introspection of Locally Cached Models
  • Added nvidia-l40s as a valid TRT compilation target

New Features

  • Workflow Profiling docs
  • Roboflow Vision Events workflow block
  • CUDA memory leak profiling script
  • SAM3 class_mapping support
  • Model registry routing metadata with GPU VRAM tracking

Full Release Notes

## What's Changed
* Add change to inform caller about model loading failures due to environment constrains violated by @PawelPeczek-Roboflow in https://github.com/roboflow/inference/pull/2180
* Add Introspection of Locally Cached Models by @yeldarby in https://github.com/roboflow/inference/pull/2161
* Pass ROBOFLOW_ENVIRONMENT env to modal based on PROJECT by @grzegorz-roboflow in https://github.com/roboflow/inference/pull/2179
* Fix missing sidebar nav for workflow blocks in Zensical docs by @Erol444 in https://github.com/roboflow/inference/pull/2182
* Add Workflow Profiling docs by @Erol444 in https://github.com/roboflow/inference/pull/2178
* Fix botocore.docs missing in jp71 image by @alexnorell in https://github.com/roboflow/inference/pull/2184
* fix(detection_event_log): use frame_timestamp for accurate absolute timestamps by @jeku46 in https://github.com/roboflow/inference/pull/2150
* Fix/windows build by @PawelPeczek-Roboflow in https://github.com/roboflow/inference/pull/2188
* Add CUDA memory leak profiling script by @hansent in https://github.com/roboflow/inference/pull/2193
* Bug/dg 314 show nicer errors webrtc by @rafel-roboflow in https://github.com/roboflow/inference/pull/2103
* Add Roboflow Vision Events workflow block by @rvirani1 in https://github.com/roboflow/inference/pull/2192
* Add routing metadata in model registry (and GPU VRAM tracking and reporting for loaded models) by @hansent in https://github.com/roboflow/inference/pull/2183
* [codex] Fix request metadata tracking for nested model-manager decorators by @hansent in https://github.com/roboflow/inference/pull/2198
* feat(sam3): add class_mapping support to SAM3 v3 block by @felipe-tomino in https://github.com/roboflow/inference/pull/2196
* Add memory backend for model monitoring cache by @hansent in https://github.com/roboflow/inference/pull/2194
* Add llms.txt and per-page markdown for LLM-friendly docs by @Erol444 in https://github.com/roboflow/inference/pull/2185
* Add nvidia-l40s as a valid TRT compilation target in `inference-cli` by @mkaic in https://github.com/roboflow/inference/pull/2121
* Bump version to `1.2.1` by @PawelPeczek-Roboflow in https://github.com/roboflow/inference/pull/2199
* Support yololite with fused NMS by @leeclemnet in https://github.com/roboflow/inference/pull/2203
* test(sam3): add tests for class_mapping feature in SAM3 v3 block by @felipe-tomino in https://github.com/roboflow/inference/pull/2202
* Remove enterprise-only flag from Vision Events block by @rvirani1 in https://github.com/roboflow/inference/pull/2207

## New Contributors
* @felipe-tomino made their first contribution in https://github.com/roboflow/inference/pull/2196

**Full Changelog**: https://github.com/roboflow/inference/compare/v1.2.0...v1.2.1