entries ai · verified 2026-07-09 · react-native

On-device AI / ML

draftedconfidence: lowearly: 1/12 of this tier graded so far (0 overturned) — the public scorecard →

re-verified 4× — 2026-07-13 · 2026-07-09 · 2026-07-07 · 2026-07-06 · changelog

recommendation

Cross-platform on-device models → react-native-executorch; Apple-only → @react-native-ai/apple; local RAG → react-native-rag. Fast-moving (confidence: low) — validate on-device perf before committing.

offline speech-to-text → expo-speech-recognition / whisper
private/offline LLM over local docs → react-native-rag

Low confidence — fast-moving or lightly-vetted domain: treat the pick as a vetted lead and prototype before committing.

Options & tradeoffs

the field considered — and why each one isn’t the default here

option	tradeoff	evidence
react-native-executorch	on-device models (Meta ExecuTorch); v0.8 adds VLM + runOnFrame worklet w/ Vision Camera; tool-calling/embeddings since 0.4, OCR since 0.3	6k/wk
@react-native-ai/apple	Apple on-device LLM + SpeechAnalyzer (iOS 26); zero cloud; RN 0.80+ (preview)
@react-native-ai/adk	Google Gemini on ANDROID via the Agent Development Kit — on-device Gemini Nano + cloud models behind one Vercel-AI-SDK provider; tool calling + streaming; same react-native-ai family as the Apple provider (v0.12)
react-native-nitro-mlx	Apple MLX on-device LLM + TTS/STT via Nitro Modules, with tool calling (ChatSession API); iOS 26+ ONLY; v0.5, ~100★ — niche/early	20/wk
react-native-rag	on-device retrieval-augmented generation over ExecuTorchLLM; local vector store	82/wk
offline speech (expo-speech-recognition / whisper)	speech-to-text without network

evidence: npm weekly downloads (signals snapshot) · “ships in n/D” = adoption across the production-app census, honest denominators

npm weekly downloads (from the corpus's last signals run): react-native-executorch 6k · @react-native-ai/apple 35k · @react-native-ai/adk 261 · react-native-rag 82 · react-native-nitro-mlx 20

Verified notes

Fast-moving area surfaced across multiple RN Rewind issues; recommend tentatively (confidence: low) and validate on-device before committing.

Canonical reading

Editorial annotations on why each piece matters — the articles themselves are the originals; read them there.

On-Device AI Beats Cloud for TTS — Here's WhyMateusz Kopciński (Software Mansion)

A reasoned cost/latency/privacy case for on-device TTS (Kokoro via react-native-executorch), with the economics and the tradeoffs (voice/language coverage, a custom C++ phonemizer). A durable framing for the on-device-vs-cloud decision.

Sources

Related in ai: ai-ui · ai-devtools