Hermes hosting
Nous Research’s Hermes models, served privately behind your own API.
Hermes hosting at AIHostBox means a dedicated, single-tenant Hermes 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 $16.99 per month and the admin account, the data and any API keys are yours.
Hermes is a family of open-weight models from Nous Research, built by fine-tuning strong base models rather than training from scratch. Hermes 3 8B sits on Meta’s Llama 3.1 8B; Hermes 4 14B sits on Qwen3 14B. Both use the ChatML prompt format and a documented tool-calling convention, where the model returns function calls as JSON inside tool_call tags. That consistency is why agent frameworks and roleplay front-ends often ship presets for Hermes specifically.
Hermes 4 adds hybrid reasoning: it can think inside explicit think tags before answering, or answer directly, depending on how you prompt it. On a laptop the model competes with everything else for memory and stops when the lid closes. On a server it becomes a fixed endpoint that several tools can share, with its weights, logs and prompts kept on hardware you rent rather than on a third party’s inference service.
Hermes 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 Hermes
One GGUF inference server loaded with a quantized Hermes build, exposing an OpenAI-compatible /v1 chat completions API on your domain with an API key.
What we configure for you
- Model file chosen for your plan: Hermes 3 Llama 3.1 8B or Hermes 4 14B at Q4
- ChatML chat template so system prompts and tool calls are formatted as Nous documents them
- Context length and thread count matched to your dedicated cores and RAM
- Sampling defaults from the model card (Hermes 4: temperature 0.6, top_p 0.95, top_k 20)
- API key on the endpoint and TLS on your own subdomain
Which plan fits
Starter (8 GB, 4 vCPU) is the plan for Hermes 3 8B and gives roughly 5-15 tokens/s at Q4, which is fine for chat and background agents. Plus (12 GB, 6 vCPU) is needed for Hermes 4 14B; it fits comfortably but generates noticeably slower than the 8B.
The honest limits
This is CPU inference: long reasoning traces and many concurrent users will queue. The larger Hermes 4 models (70B, 405B) are out of reach without a GPU plan.
What people run Hermes for
Agent backbone
A model that respects a long system prompt without drifting.
Private assistant
Run the model behind your own endpoint with no third-party logging.
Tool-calling experiments
Test function calling against a model whose tool format is published and stable.
Character and roleplay back-end
Serve a steerable model to a front-end such as SillyTavern over the private network.
Licence and version policy
| Item | Position |
|---|---|
| Upstream licence | Model weights under their own terms; server MIT |
| 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 Hermes
Open WebUI
A ChatGPT-style workspace for your team, pointed at any model you like.
from $5.99/moDetails →
SillyTavern
The character-driven chat front-end, online wherever you are.
from $5.99/moDetails →
Letta
Stateful agents with long-term memory that survives the context window.
from $11.99/moDetails →
LiteLLM
One OpenAI-compatible endpoint in front of every model provider you use.
from $5.99/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 →
LocalAI
A drop-in OpenAI replacement for text, images, audio and embeddings.
from $16.99/moDetails →
KoboldCpp
A single-binary GGUF server with its own UI and a KoboldAI API.
from $13.99/moDetails →
llama.cpp
The reference GGUF inference server, with nothing between it and the CPU.
from $13.99/moDetails →Hermes hosting: your questions
Which Hermes version do you run?
Is it as fast as a GPU?
Which Hermes models fit on each plan?
Does tool calling work through the API?
Can I turn Hermes 4’s reasoning on and off?
What licence do the weights carry?
Run Hermes without running a server
A dedicated Hermes instance, deployed and maintained by us. Month to month, cancel whenever.