# Tiiny-AI/PowerInfer

High-speed Large Language Model Serving for Local Deployment

Repository: https://github.com/Tiiny-AI/PowerInfer
Canonical: https://ross.abutalabs.com/products/powerinfer
Language: C++
License: MIT
License Family: permissive
Topics: large-language-models, llama, llm, llm-inference, local-inference
Last push: 2026-05-11T06:48:06+00:00

## Health v2 (maintenance only)
Score: 63/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 81, release rhythm 35, longevity 70
- inputs: {"age_days": 993, "days_push": 114, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 9752, forks 595 (observed 2026-08-28T04:10:37.054880+00:00)

## What it is
PowerInfer is a CPU/GPU hybrid LLM inference engine that exploits activation locality to run large language models fast on consumer-grade hardware, including desktops and smartphones. It supports sparse models like TurboSparse and ProSparse and is MIT-licensed C++ software.

## Use cases
- run an LLM locally on a consumer GPU
- serve large language models on a single desktop machine
- speed up inference of sparse ReLU-activated LLMs
- run LLM inference on a smartphone
- deploy Llama-style models offline on my own hardware

## When to choose
- you want fast local LLM inference on consumer-grade CPU/GPU hardware
- you are using sparsified models like TurboSparse or ProSparse
- you need on-device inference on smartphones or edge devices

## When to avoid
- you need multi-node distributed serving at datacenter scale
- you want a managed cloud inference API
- your models are not compatible with its supported architectures

## Facets
- artifact type: library
- maturity: active
- function: llm-inference, gpu-computing, machine-learning
- domain: large-language-models, machine-learning, gpu-computing
- platform: windows, cpp
- tags: cpu-gpu-hybrid, activation-locality, sparse-models, local-deployment, consumer-gpu, on-device-inference, linux, macos, gpu, android

## Member repositories
- Tiiny-AI/PowerInfer (main) score 63

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:37.054880+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 2026-08-29T17:21:12.047785+00:00, confidence not recorded.
  - readme: https://github.com/Tiiny-AI/PowerInfer (fetched 2026-08-28T04:10:37.054880+00:00, sha ab468dfeb814)
- Data as of 2026-08-30T08:39:29.467469+00:00.