QVAC vs KoboldCpp
QVAC
Open-source ecosystem for local-first peer-to-peer AI on every platform.
KoboldCpp
One self-contained binary covering text, images, video, music, speech recognition and speech synthesis.
Key differences
KoboldCpp ships as a single self-contained binary covering text generation, Stable Diffusion and Flux image generation, Whisper transcription, several speech synthesis engines, video through WAN 2.2 and music through Ace Step. Its modality coverage overlaps QVAC's and extends past it into video and music.
The two are packaged for different uses. KoboldCpp is an executable a person runs, with a web interface and an OpenAI-compatible endpoint attached. QVAC is a library linked into an application, with a typed API and no interface of its own.
The licences differ in scope. KoboldCpp is AGPL-3.0, a copyleft licence whose obligations extend to software offered over a network and which requires source for the combined work. The QVAC SDK is Apache 2.0 and the Fabric engine is MIT. For a closed-source commercial product that distinction is a legal constraint, and both licences repay reading in full.
This page compares KoboldCpp v1.120, released 29 August 2026, against QVAC 0.18.2, meaning the SDK together with the Fabric inference engine at v10297.1.1. Every row was checked against the project's own documentation and release notes on 4 September 2026. Both projects move quickly, so check the current release before you make a decision on either one.
Feature matrix
Feature
QVAC
KoboldCpp
PLATFORMS
macOS
YesYesWindows
YesYesLinux
YesYesAndroid
YesVia Termux
iOS
YesAI TASKS
Text generation
YesYesTranscription
YesYesTranslation
YesImage generation
YesYesOCR
YesText-to-speech
YesYesRUNTIME SUPPORT
Node.js
YesVia HTTP
Bare
YesExpo
YesHTTP server
YesYesCLI
YesYesP2P
Peer discovery
YesInference delegation
YesEncrypted transport
YesMOBILE SUPPORT
On-device inference
YesVia Termux
LoRA fine-tuning on mobile
YesMobile SDK
YesLICENSING
License
Apache 2.0
AGPL-3.0
Open weights tooling
YesYesWhen to choose QVAC
You need a permissive licence for a commercial product.
You are embedding inference in an application rather than running a tool.
You target iOS, or Android without Termux.
You need typed APIs and on-device fine-tuning.
When to choose KoboldCpp
You want every modality from one file today.
AGPL-3.0 fits your project.
You want an interface included rather than written.
You are generating images, video or music locally for yourself.
Compare
Ready to build with QVAC?
One SDK, every platform, no rent. Grab it and ship your own local-first AI.
npm install @qvac/sdk