Introducing the Qtum AI Router

If you're building AI agents or running personal assistants locally, you know the drill: you want GPT-5.4-mini for cheap reasoning, DeepSeek-v4-pro for fast chat workloads, Kimi-k2.6 for that massive 200k context window, and MiniMax-M2.5 for multimodal tasks.

https://qtum.ai

But managing multiple accounts, tracking fragmented balances across half a dozen providers, and dealing with manual endpoint switching is a massive pain.

We just put together an integration that fixes this. The Qtum AI Router just hit beta, acting as a unified inference layer that supports both OpenAI Chat Completions and Anthropic Messages wire formats out of the box. To show it off, we wrote an open-source plugin for OpenClaw (the personal AI assistant gateway) so you can tap into the entire model catalog using a single API key and centralized billing (settled in QTUM credits).

🔄 The Difference: With vs. Without a Router

Feature	Without a Router	With Qtum Router
Accounts	❌ Multiple accounts	One account
API Keys	❌ Multiple API keys	One API key
Balances	❌ Multiple balances	One balance
Endpoints	❌ Different endpoints	One endpoint
Switching	❌ Manual provider switching	Simple model switching

🧠 The Current Unified Catalog

Every model listed below is accessible via a single key, with input/output pricing settled seamlessly in QTUM credits:

GPT Family: gpt-5.5, gpt-5.4, gpt-5.4-mini (Vision supported)
DeepSeek Family: deepseek-v4-pro, deepseek-v4-flash (Low-cost text specialists)
Qwen Family: qwen3.7-max, qwen3.6-plus, qwen3.6-max-preview, qwen3.6-flash (Multilingual)
Kimi Family: kimi-k2.6 (200k long-context)
GLM Family: glm-5.1 (Vision)
MiniMax Family: MiniMax-M2.5 (Frontier multimodal)

🛠️ The Onboarding & Setup

If you don't have OpenClaw installed yet, you can spin up a healthy gateway process, a default agent, and an onboarding wizard on your Linux server in 3 quick commands:

Bash
# 1. Install via npm (Node 20+ required)
npm i -g openclaw
# 2. Run the onboarding wizard to pick ports, generate tokens, and setup systemd 
openclaw onboard
# 3. Confirm the gateway is up
openclaw status

Once your status checks out, here is the quick 3-step breakdown to wire in the Qtum backend.

1. Install the Qtum Plugin

The plugin registers qtum as a first-class provider in OpenClaw. Run this block to clone the repo, install the extension path, and restart your user service:

Bash
sudo -iu openclaw bash <<'EOF' 
export PATH=$HOME/.npm-global/bin:$PATH 
export XDG_RUNTIME_DIR=/run/user/$(id -u) 
export DBUS_SESSION_BUS_ADDRESS=unix:path=/run/user/$(id -u)/bus 

cd ~

git clone https://github.com/Dannnymack/openclaw-plugin-qtum

openclaw plugins install ./openclaw-plugin-qtum

systemctl --user restart openclaw-gateway

EOF

(You can verify the extension environment loaded successfully by running openclaw plugins doctor)

2. Add Your API Key

Grab your unified credential from qtum.ai and drop it into your main agent's auth-profiles.json. This tells OpenClaw to pipe everything through the single provider backend:

JSON

{

"version":1,

"profiles":{

 "qtum:default":{

  "type":"api_key",

  "provider": "qtum",

  "key":"sk-..your-key..."

 }

},

"lastGood":{

 "qtum":"qtum:default"

 }

}

3. Switch Models Globally on the Fly

Now you can swap the brains of your agent dynamically with a single configuration flag. No code modifications required:

For cheap text workloads: openclaw models set qtum/deepseek-v4-pro
For reasoning & vision: openclaw models set qtum/gpt-5.4-mini
For frontier multimodal tasks: openclaw models set qtum/MiniMax-M2.5

Just hit it with a quick systemctl --user restart openclaw-gateway to apply your model switch!

🌐 Why Route Through a Blockchain Layer?

Zero Vendor Lock-In: Because it speaks standard OpenAI and Anthropic API schemas, your client code doesn't change. Just point it at the new base URL and walk away whenever you want. There's no need to purchase token "bundles" or sign up for recurring subscriptions. You simply pay for what you use. All you need is a MetaMask wallet or a Google login to access the Qtum AI Router.
Underlying Network Resiliency: This isn't just a fragile web proxy script. It’s backed by a Proof-of-Stake blockchain network running production traffic with zero downtime since 2017, combining Bitcoin’s UTXO security model with EVM smart contracts. Paying with cryptocurrency allows users who don't have access to a credit card, or simply don't want to use a credit card, to get access to tokens.
Auto-Updating Catalog: The router updates live. When a new model drops, the gateway fetches the updated manifest on a restart—no client tool updates required.

Would love to get some feedback from the community on this integration! If you want to check out the metrics dashboard, monitor request volumes, or view pricing per million tokens, head over to console.qtum.ai.