# huggingface/candle

Minimalist ML framework for Rust

Repository: https://github.com/huggingface/candle
Canonical: https://ross.abutalabs.com/products/candle
Language: Rust
License: Apache-2.0
License Family: permissive
Last push: 2026-08-23T13:19:46+00:00

## Health v2 (maintenance only)
Score: 73/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 35, longevity 83
- inputs: {"age_days": 1171, "days_push": 10, "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 20955, forks 1725 (observed 2026-08-28T04:11:30.532753+00:00)

## What it is
Candle is a minimalist machine learning framework for Rust focused on performance and ease of use, with CPU and CUDA GPU support. It ships tensor operations and ready-made examples for models like LLaMA, Whisper, YOLO, and Segment Anything, including browser-based WASM demos.

## Use cases
- run LLM inference in Rust
- tensor math with GPU acceleration in Rust
- run whisper speech recognition locally
- deploy ML models to wasm in the browser
- run yolo object detection without python
- image segmentation with segment anything in rust

## When to choose
- you want a lightweight, dependency-light ML stack in pure Rust
- you need serverless or WASM deployment of ML models
- you want to avoid Python runtime overhead for inference

## When to avoid
- you need the full PyTorch ecosystem, training tooling, or exotic ops
- you rely on Python libraries or pretrained pipelines out of the box
- you need broad hardware support beyond CUDA/CPU

## Facets
- artifact type: framework
- maturity: active
- function: machine-learning, deep-learning, llm-inference, speech-recognition, computer-vision, gpu-computing, wasm
- domain: machine-learning, deep-learning, large-language-models, computer-vision, speech-processing, developer-tools
- platform: rust, wasm, cross-platform, cli
- tags: tensor-library, rust-ml, huggingface, cuda, serverless-inference, gpu

## Member repositories
- huggingface/candle (main) score 73

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:30.532753+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 2026-08-29T16:58:30.961631+00:00, confidence not recorded.
  - readme: https://github.com/huggingface/candle (fetched 2026-08-28T04:11:30.532753+00:00, sha b5c646b6b29f)
- Data as of 2026-08-30T08:39:29.467469+00:00.