v0.10.0-libs-1.4.1 (ExecuTorch 1.4.1)

Native lib artifacts for react-native-executorch built against ExecuTorch 1.4.1.

Same artifact set and layout as v0.10.0-libs, which stays on ExecuTorch 1.3.1 and is unchanged. This tag exists so the two native surfaces can be tested side by side.

Built from

• ExecuTorch: software-mansion-labs/executorch @ @ms/separate-backends-1.4.1 (226b4714)
• Tokenizers: software-mansion-labs/pytorch-tokenizers @ @ms/build-1.4.1 (a03231a)

Contents

Platform	Artifacts
Android arm64-v8a	core, xnnpack, vulkan, opencv
Android x86_64	core, xnnpack, vulkan, opencv
iOS	core (ExecutorchLib.xcframework), xnnpack, coreml, mlx (device slice + mlx.metallib)
Shared	headers

OpenCV tarballs are carried over unchanged from v0.10.0-libs; OpenCV does not depend on ExecuTorch.

Using it

nativeLibsVersion is still 0.10.0, so this release is not resolved by default. Point the downloader at it explicitly, and give it a separate cache directory:

RNET_BASE_URL=https://github.com/software-mansion/react-native-executorch/releases/download/v0.10.0-libs-1.4.1 \
RNET_LIBS_CACHE_DIR=~/.cache/react-native-executorch/1.4.1 \
INIT_CWD=<repo-root> node scripts/download-libs.js

The RNET_LIBS_CACHE_DIR override is required, not optional. The default cache directory is keyed on nativeLibsVersion alone, so without it these artifacts would share ~/.cache/react-native-executorch/0.10.0 with the 1.3.1 tarballs and a warm cache would silently win.

Status

Pre-release, for verification only. Android demo apps (CV, NLP, speech) have been run against these artifacts on device; iOS verification is in progress. Whisper MLX in live-streaming mode is a known open item under investigation and is not yet confirmed either way against 1.3.1.

v0.9.3

What's Changed

• feat: added system and bundleId to telemetry to version 0.9 by @mkopcins in #1311
• fix(resource-fetcher): avoid this in static fs field initializer by @danchily2 in #1310
• fix(android): forward ExecuTorch consumer ProGuard rules by @injunchoi98 in #1325

Full Changelog: v0.9.2...v0.9.3

v0.8.5

What's Changed

• feat: added system and bundleId to telemetry to version 0.8 by @mkopcins in #1312

Full Changelog: v0.8.4...v0.8.5

v0.10.0-libs

Prebuilt native lib artifacts for react-native-executorch, pinned via nativeLibsVersion in package.json (see download-libs.js).

Pre-release while the native surface for 0.10.0 is still settling; will be promoted to a full release (--prerelease=false) once final. Do not re-upload assets on this tag — cut a new v*-libs version for native fixes.

rne-rewrite on-demand libs (test)

Test artifacts for PR #1283 (rne-rewrite on-demand native libs). Fresh Android and iOS split-backend build (ET 1.3.1 @ms/separate-backends). Will be deleted.

v0.9.2

What's Changed

• RF-Detr keypoint preview support
• LLM fix: snapshot vision encoder output before caching
• Speed up topp sampling

Full Changelog: v0.9.1...v0.9.2

v0.9.1

What's Changed

• Addition of Gemma 4 with Vulkan, MLX and Xnnpack support
• Whisper improvements on iOS - 50% smaller, 30% faster due to switch from fp32 to fp16

Full Changelog: v0.9.0...v0.9.1

v0.9.0

React Native ExecuTorch v0.9.0 🔒 🚀

This release, as always, brings a lot of fresh goodies to the React Native world ⚛️. It introduces new things across the stack, whether that's speech, vision, or NLP, check it out 👇🏻

What's new?

Speech

• Up to 10x performance improvements for transcription - We introduced a way to integrate VAD into Whisper transcription pipeline, along with underlying algorithm optimizations & NPU model exports, resulting in significantly improved performance docs
• Multilingual Text-to-Speech - We introduced a lot of new languages to the TTS API, including 🇵🇱 🇪🇸 🇮🇹 🇫🇷 🇮🇳 docs
• Text-to-Speech improvements - We improved the phonemization mechanisms to significantly enhance the pronunciation quality for existing languages
• VAD streaming API - You can now run voice activity detection in continuous, stream fashion docs

NLP

• OpenAI's Privacy Filter - A new usePrivacyFilter API, making it easy to detect private user information from text, fully locally docs
• Multilingual text embeddings - It is now possible to generate embeddings across different languages docs
• Qwen 3.5 support - Another LLM in our collection docs
• LFM2.5-VL-450M (quantized) & LFM2.5 350M - New multimodal and text-only LLMs from Liquid AI added to the registry
  docs
• Smarter LLM sampling - min_p and repetition_penalty sampling, per-model defaults, and letterboxed multimodal image preprocessing docs

Computer vision

• Pose estimation - A new Pose Estimation API with YOLO 26 docs
• FastSAM with prompts - Segment Anything style masks driven by point, box, or text prompts, fully on-device docs

General

• ExecuTorch runtime bumped to v1.2.0 - Updated native binaries and headers across iOS and Android, including the new AlreadyLoaded error mapping docs
• Typed models registry - A new declarative models accessor in constants/modelRegistry.ts, grouped one-to-one with hooks, so you can pick a backend / precision without juggling URL constants docs
• Extended tokenizer support - Rebuilt tokenizer binaries with unigram & word-level support, plus previously unsupported pre-tokenizers, decoders, and post-processors
• Graceful native-lib degradation on Android - Apps no longer crash at startup on 32-bit / unsupported ABIs; isAvailable === false lets you render a fallback UI instead docs
• iOS deployment target raised to 17.0 - Matches the prebuilt ExecuTorch libraries; apps targeting older iOS versions will need to bump

Breaking changes ⚠️

• TextToImage returns a file:// URI instead of base64 — useTextToImage().generate() (and TextToImageModule.forward) now resolves to a path on disk pointing to a PNG, instead of a base64-encoded payload. The pngjs dependency is no longer needed.
• Multilingual TTS API reshape — The Kokoro pipeline now takes per-language voice / voice-quality configuration to support voice cloning and fine-tuned variants. Existing TTS call sites need to migrate.
• Speech-to-text quantized model constants removed — Predefined constants for quantized Whisper variants were dropped; the quality delta vs. the originals didn't justify the extra API surface. iOS now defaults to the new CoreML-backed Whisper builds.
• OCRDetection.bbox is now Bbox, not Point[] — The native vision stack moved to a unified Point struct everywhere; consumers reading raw bbox geometry off OCR results need to update.
• ResourceFetcherAdapter.fetch return shape changed — Custom adapters must now return { path: string; wasDownloaded: boolean }[] instead of string[], so callers can tell which resources were freshly downloaded vs. served from cache. Update any custom adapter implementations accordingly. docs

v0.8.4

What's Changed

• chore: Release v0.8.4 by @msluszniak in #1108

Full Changelog: v0.8.3...v0.8.4

v0.8.3

What's Changed

• chore: Update GitHub Actions and bump Node to 24 across all workflows… by @mkopcins in #1071
• chore: Release v0.8.3 by @msluszniak in #1072

Full Changelog: v0.8.2...v0.8.3