# ngxson/wllama

WebAssembly binding for llama.cpp - Enabling on-browser LLM inference

Repository: https://github.com/ngxson/wllama
Canonical: https://ross.abutalabs.com/products/wllama
Homepage: https://huggingface.co/spaces/ngxson/wllama
Language: TypeScript
License: MIT
License Family: permissive
Topics: llama, llamacpp, llm, wasm, webassembly
Last push: 2026-09-03T00:42:06+00:00

## Health v2 (maintenance only)
Score: 92/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 100, release rhythm 99, longevity 64
- inputs: {"age_days": 903, "days_push": 0, "days_rel": 6, "gap_med": 11.0, "n_releases_24m": 35}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1191, forks 119 (observed 2026-09-03T02:15:15.528613+00:00)

## What it is
wllama is a WebAssembly binding for llama.cpp that enables running LLM inference entirely in the browser, with WebGPU support, multimodal input, and tool calling. It ships as an npm package with an OpenAI-compatible typed API and runs inference in a web worker without any backend.

## Use cases
- run LLM inference in the browser without a server
- generate text completions client-side with llama.cpp models
- compute embeddings in the browser for semantic search
- build a private chatbot that keeps data on-device
- run vision models on images directly in a web app
- add tool calling to an in-browser AI assistant

## When to choose
- you want zero-backend, privacy-preserving LLM inference in a web app
- you need embeddings or completions computed client-side
- you want an OpenAI-compatible API over llama.cpp in the browser

## When to avoid
- you need to run models larger than 2GB without splitting them
- you need high-throughput server-side inference with full GPU utilization
- your users' browsers lack WebAssembly SIMD or WebGPU support

## Facets
- artifact type: library
- maturity: active
- function: llm-inference, sdk, wasm, chatbot, rag
- domain: large-language-models, artificial-intelligence, web-development, frontend
- platform: browser, wasm
- tags: llama-cpp, webgpu, in-browser-inference, openai-compatible-api, multimodal, tool-calling, embeddings, typescript, web-server, nodejs

## Member repositories
- ngxson/wllama (main) score 92

## Provenance
- Observed fields: from GitHub, fetched 2026-09-03T02:15:15.528613+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 2026-08-30T06:26:35.689295+00:00, confidence not recorded.
  - readme: https://github.com/ngxson/wllama (fetched 2026-09-03T02:15:15.528613+00:00, sha ad9773a66077)
  - homepage: https://huggingface.co/spaces/ngxson/wllama (fetched 2026-08-29T12:32:57.763986+00:00, sha 3fda121c51f5)
- Data as of 2026-08-30T08:39:29.467469+00:00.