KoboldCpp hosting
A single-binary GGUF server with its own UI and a KoboldAI API.
KoboldCpp hosting at AIHostBox means a dedicated, single-tenant KoboldCpp instance on our infrastructure: we install it, pin it to a reviewed release, put it behind TLS on your domain, back it up nightly and keep it patched. Plans start at $13.99 per month and the admin account, the data and any API keys are yours.
KoboldCpp is a single-file model server maintained by LostRuins (concedo) that wraps llama.cpp with the KoboldAI Lite interface. You point it at a GGUF file and get three things at once: a browser UI for chat, adventure and story writing, the KoboldAI API that role-play front-ends expect, and an OpenAI-compatible /v1 endpoint plus an Ollama-style API for everything else. Context shifting lets long sessions continue without reprocessing the whole prompt, which is the reason writers and character-chat users keep choosing it over plainer servers.
It also bundles optional extras in the same binary: image generation, Whisper transcription, text-to-speech, embeddings and vision through an mmproj file. On a laptop it only works while the lid is open and the fan is spinning. On a server it becomes an endpoint that SillyTavern, a phone or a script can reach from anywhere, protected by an API key and running the same model all day without tying up your own machine.
KoboldCpp hosting plans
Every plan is a dedicated single-tenant instance with NVMe storage, nightly off-box backups, TLS and managed patching. Quarterly, semi-annual and annual terms are discounted at checkout. Model tokens are not included — you bring your own key.
How we run KoboldCpp
We run the upstream KoboldCpp release as a single process on port 5001 with one GGUF model loaded from local NVMe storage; there is no database, and settings live in a .kcpps config file.
What we configure for you
- --model pointing at your chosen GGUF file, downloaded or uploaded over SFTP
- --password (KCPP_PASSWORD) so every text endpoint needs your API key
- --threads matched to the plan’s vCPUs and --contextsize matched to its RAM
- --multiuser so several clients, such as SillyTavern and a script, can queue requests
- A saved .kcpps config so the same settings come back after every restart or update
Which plan fits
Starter (8 GB, 4 vCPU) fits a 7-8B GGUF at Q4 with a moderate context, typically 5-15 tokens per second on CPU. Plus (12 GB, 6 vCPU) fits a 13B Q4 model, which reads better for fiction but generates more slowly. Larger models, or long contexts on top of 13B, need more RAM than either plan offers.
The honest limits
Everything runs on CPU, so the image generation, speech and video extras are possible but slow enough that we do not recommend them on these plans. KoboldCpp’s own help notes that --password does not protect the image endpoints.
What people run KoboldCpp for
SillyTavern backend
The most common local pairing for character chat.
Long-form generation
Context shifting keeps long sessions coherent without a full re-prompt.
Private OpenAI-compatible endpoint
Point any tool that accepts a custom base URL at /v1 and keep prompts off third-party APIs.
Interactive fiction
Use the built-in story and adventure modes in KoboldAI Lite without installing a separate front-end.
Licence and version policy
| Item | Position |
|---|---|
| Upstream licence | AGPL-3.0 |
| Sold as managed hosting | Yes |
| Tenancy | Single tenant — one customer per instance |
| Patching | Critical advisories within 24 hours, high severity within 72 |
Our full position on licences, including the applications we decline to sell, is on the licensing page.
Works well with KoboldCpp
SillyTavern
The character-driven chat front-end, online wherever you are.
from $5.99/moDetails →
Open WebUI
A ChatGPT-style workspace for your team, pointed at any model you like.
from $5.99/moDetails →
LobeChat
A fast, polished chat UI with a large plugin and assistant marketplace.
from $11.99/moDetails →More LLM runtimes we host
Ollama
Your private LLM server — pull a model, get an API, keep every token on your own box.
from $16.99/moDetails →
Hermes
Nous Research’s Hermes models, served privately behind your own API.
from $16.99/moDetails →
LocalAI
A drop-in OpenAI replacement for text, images, audio and embeddings.
from $16.99/moDetails →
llama.cpp
The reference GGUF inference server, with nothing between it and the CPU.
from $13.99/moDetails →KoboldCpp hosting: your questions
Which models can I run?
How do I stop strangers using my instance?
Can other apps use it like the OpenAI API?
How long a context can I use?
Is AGPL-3.0 a problem for me?
Run KoboldCpp without running a server
A dedicated KoboldCpp instance, deployed and maintained by us. Month to month, cancel whenever.