Wikitopia
llama.cpp
productllm_inference
Try in Playground →
Overview
Developed byGeorgi Gerganov
Founded2023
LicenseMIT License
Open source✓ Open Source
Primary languageC++
Use caseRunning LLM inference on CPU
Technical
Protocols
GGML format
Integrates with
CUDA
Also see
Alternative to
Hugging Face TransformersPyTorch inference
Based onC++
Knowledge graph stats
Claims43
Avg confidence93%
Avg freshness99%
Last updatedUpdated yesterday
WikidataQ125998452
Trust distribution
100% unverified
Governance
Not assessed
Contribute governance data →

llama.cpp

product

MIT C/C++ library for LLM inference with GGUF support, runs on CPU and GPU

Compare with...

based on

ValueTrustConfidenceFreshnessSources
C++○UnverifiedHighFresh1
C++ programming language○UnverifiedHighFresh1
GGML tensor library○UnverifiedHighFresh1
GGML library○UnverifiedModerateFresh1

supports model

ValueTrustConfidenceFreshnessSources
LLaMA○UnverifiedHighFresh1
LLaMA models○UnverifiedHighFresh1
Alpaca models○UnverifiedHighFresh1
Vicuna models○UnverifiedHighFresh1
Alpaca○UnverifiedHighFresh1
Vicuna○UnverifiedModerateFresh1
Code Llama○UnverifiedModerateFresh1
GPT4All○UnverifiedModerateFresh1

programming language

ValueTrustConfidenceFreshnessSources
C++○UnverifiedHighFresh1

open source

ValueTrustConfidenceFreshnessSources
true○UnverifiedHighFresh1

pricing model

ValueTrustConfidenceFreshnessSources
free○UnverifiedHighFresh1

integrates with

ValueTrustConfidenceFreshnessSources
CUDA○UnverifiedHighFresh1

primary use case

ValueTrustConfidenceFreshnessSources
Running LLM inference on CPU○UnverifiedHighFresh1
local inference of large language models○UnverifiedHighFresh1
CPU inference for LLaMA models○UnverifiedHighFresh1
LLM inference on CPU○UnverifiedHighFresh1
CPU-based LLM inference○UnverifiedHighFresh1
Local LLM inference with minimal dependencies○UnverifiedHighFresh1

supports quantization

ValueTrustConfidenceFreshnessSources
true○UnverifiedHighFresh1
4-bit quantization○UnverifiedHighFresh1

supports platform

ValueTrustConfidenceFreshnessSources
Linux○UnverifiedHighFresh1
macOS○UnverifiedHighFresh1
Windows○UnverifiedHighFresh1

quantization support

ValueTrustConfidenceFreshnessSources
GGML format○UnverifiedHighFresh1

requires

ValueTrustConfidenceFreshnessSources
CPU○UnverifiedHighFresh1
C++ compiler○UnverifiedHighFresh1
quantized model weights○UnverifiedHighFresh1
no GPU○UnverifiedHighFresh1

platform support

ValueTrustConfidenceFreshnessSources
cross-platform○UnverifiedHighFresh1

maintained by

ValueTrustConfidenceFreshnessSources
Georgi Gerganov○UnverifiedHighFresh1

developed by

ValueTrustConfidenceFreshnessSources
Georgi Gerganov○UnverifiedHighFresh1

license type

ValueTrustConfidenceFreshnessSources
MIT License○UnverifiedHighFresh1

supports protocol

ValueTrustConfidenceFreshnessSources
GGML format○UnverifiedHighFresh1

supports hardware

ValueTrustConfidenceFreshnessSources
CUDA GPUs○UnverifiedHighFresh1
Apple Silicon○UnverifiedHighFresh1

uses quantization

ValueTrustConfidenceFreshnessSources
4-bit and 8-bit○UnverifiedHighFresh1

founded year

ValueTrustConfidenceFreshnessSources
2023○UnverifiedModerateFresh1

alternative to

ValueTrustConfidenceFreshnessSources
Hugging Face Transformers○UnverifiedModerateFresh1
PyTorch inference○UnverifiedModerateFresh1

Alternatives & Similar Tools

Hugging Face Transformers
alternative to
Compare →
PyTorch inference
alternative to
Compare →

Commonly Used With

CUDA

Related entities

Claim count: 43Last updated: 4/9/2026Edit history