# google-ai-edge/LiteRT-LM

LiteRT-LM is Google's production-ready, high-performance, open-source inference framework for deploying Large Language Models on edge devices.

Repository: https://github.com/google-ai-edge/LiteRT-LM
Canonical: https://ross.abutalabs.com/products/litert-lm
Homepage: https://ai.google.dev/edge/litert-lm
Language: C++
License: Apache-2.0
License Family: permissive
Topics: edge-ai, on-device-ai, on-device-llm
Last push: 2026-08-26T23:42:18+00:00

## Health v2 (maintenance only)
Score: 86/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 98, longevity 36
- inputs: {"age_days": 506, "days_push": 7, "days_rel": 15, "gap_med": 12.5, "n_releases_24m": 15}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 6298, forks 697 (observed 2026-08-28T04:09:41.497471+00:00)

## What it is
LiteRT-LM is Google's production-ready, high-performance open-source framework for running large language models on edge devices, built as an orchestration layer over LiteRT. It provides C, Python, Swift, JavaScript, Kotlin, and Flutter APIs with GPU/NPU acceleration across Android, iOS, Web, desktop, and IoT platforms.

## Use cases
- run llms on-device on android or ios
- deploy gemma locally on a raspberry pi
- run a local llm in the browser with webgpu
- integrate on-device ai into a flutter app
- run llm inference with gpu or npu acceleration
- build an offline chatbot without a server
- add function calling to an on-device ai agent

## When to choose
- you need production-grade on-device LLM inference across mobile, web, and desktop
- you want hardware-accelerated (GPU/NPU) inference on edge devices
- you need multimodal inputs (vision, audio) and tool use locally
- you want to run Gemma, Llama, Phi-4, or Qwen models offline

## When to avoid
- you need server-scale inference with large models on datacenter GPUs
- you only need cloud API access to LLMs without local deployment
- you need fine-tuning or training rather than inference
- you require a model format other than LiteRT/.litertlm

## Facets
- artifact type: library
- maturity: active
- function: llm-inference, machine-learning, sdk, cli, gpu-computing
- domain: large-language-models, artificial-intelligence, mobile-development, cross-platform, embedded-systems
- platform: cross-platform, python, cpp, cli, embedded, windows
- tags: edge-ai, on-device-llm, litert, gemma, hardware-acceleration, multimodal, function-calling, webgpu, raspberry-pi, android, ios, web-server, gpu, macos, linux

## Member repositories
- google-ai-edge/LiteRT-LM (main) score 86

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:41.497471+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 2026-08-29T17:46:16.185019+00:00, confidence not recorded.
  - readme: https://github.com/google-ai-edge/LiteRT-LM (fetched 2026-08-28T04:09:41.497471+00:00, sha e8fb9b55880e)
  - homepage: https://ai.google.dev/edge/litert-lm (fetched 2026-08-29T08:42:46.795742+00:00, sha ee0397567986)
- Data as of 2026-08-30T08:39:29.467469+00:00.