# RunanywhereAI/runanywhere-sdks

Production ready toolkit to run AI locally

Repository: https://github.com/RunanywhereAI/runanywhere-sdks
Canonical: https://ross.abutalabs.com/products/runanywhere-sdks
Homepage: https://www.runanywhere.ai
Language: C++
License: NOASSERTION
License Family: other
Topics: ios, kotlin, llm, swift, inference, multimodal, on-device-ai, voice-ai, android, edge, llamacpp, apple-intelligence, ollama, cpp, diffusion-models, flutter, react-native, vlm, web, websdk
Last push: 2026-08-27T00:06:24+00:00

## Health v2 (maintenance only)
Score: 85/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 99, longevity 29
- inputs: {"age_days": 408, "days_push": 7, "days_rel": 7, "gap_med": 1, "n_releases_24m": 20}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 10282, forks 369 (observed 2026-08-28T04:10:41.465990+00:00)

## What it is
RunAnywhere is a set of cross-platform SDKs (Swift, Kotlin, React Native, Flutter, TypeScript, C++) over a shared C++ core for running AI models fully on-device — LLM chat, speech-to-text, text-to-speech, vision, embeddings, RAG, and image generation. It routes inference to hardware-appropriate engines (llama.cpp, MLX, Core ML, Qualcomm NPU, WebGPU) with a unified API and a console for model deployment and telemetry.

## Use cases
- run llm inference locally on ios and android
- build an offline voice assistant with stt and tts
- add private on-device chatbot to a mobile app
- run speech-to-text without a server
- generate images on-device with stable diffusion
- do rag with local embeddings on mobile
- run ai models in the browser with webgpu
- deploy on-device ai across a fleet of devices

## When to choose
- you need private, offline AI inference on phones, browsers, or desktops
- you want one unified API across multiple platforms and model engines
- you need low-latency voice pipelines (STT, TTS, VAD) on mobile
- you want to exploit device NPUs/GPUs like Apple Neural Engine or Qualcomm Hexagon

## When to avoid
- you need large-scale server-side inference on datacenter GPUs
- you require a permissive open-source license (the license is custom, not OSI-approved)
- you only need a simple cloud API call with no on-device footprint
- your target platform is not among the supported SDKs

## Facets
- artifact type: library
- maturity: active
- function: llm-inference, speech-recognition, tts, rag, machine-learning, sdk, image-processing, chatbot
- domain: artificial-intelligence, large-language-models, mobile-development, cross-platform, speech-processing, privacy
- platform: windows, browser, wasm, cpp, python, cross-platform
- tags: on-device-ai, edge-ai, llamacpp, npu-acceleration, voice-agents, multimodal, offline-inference, react-native, flutter, mobile-sdk, ios, android, macos, linux, web, swift

## Member repositories
- RunanywhereAI/runanywhere-sdks (main) score 85

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:41.465990+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:19:20.706563+00:00, confidence not recorded.
  - readme: https://github.com/RunanywhereAI/runanywhere-sdks (fetched 2026-08-28T04:10:41.465990+00:00, sha 83b4f239c66f)
  - homepage: https://www.runanywhere.ai (fetched 2026-08-29T08:18:42.258767+00:00, sha 85893fa2b93d)
  - site_page: https://www.runanywhere.ai/about (fetched 2026-08-29T08:18:42.268674+00:00, sha acbe33776b25)
  - site_page: https://docs.runanywhere.ai (fetched 2026-08-29T08:18:42.279881+00:00, sha 557d2f7aa1a7)
- Data as of 2026-08-30T08:39:29.467469+00:00.