AI Gateway
Every model your business uses, under one set of rules.
Choose which models do the work, run them on your own provider account or on ours, and give people, agents and your own code a budget they cannot overrun.
One gateway between your business and every model.
Your team in the chat, your agents and Minis, and your developers' code all reach the models through the AI Gateway, and its rules hold for each of them.
Browse the models- Your team, in the chat
- Agents and Minis
- Your own code, through the API
- Which models may run
- Whose key pays
- Who may spend how much
- No prompts or answers kept
- Anthropic
- OpenAI
- Meta
- Mistral
- DeepSeek
- xAI
- Alibaba
- Cohere
- Amazon
- NVIDIA
- Moonshot AI
Pick the model for the job. In the chat, choose which model answers: Claude, GPT, Gemini, Llama, Mistral, DeepSeek and more. An admin sets the default for everyone and which providers and models may run.
Your account, your bill. Add your organization's key for Anthropic, OpenAI, Google or DeepSeek, and their models run on your account, paid to your provider directly. Without a key, everything runs on SalexHub, paid in AI credits.
One gateway for your own code. Your developers call the same models through an OpenAI-compatible API, with keys of its own, under the same rules and the same bill.
Budgets and privacy, set once. Give every person, agent, workflow and API key a monthly limit, and see who spent what. Providers keep no prompts or answers by default, and card numbers, IBANs and keys are masked before a model reads them.
The same models, from your code.
An Owner or Admin creates an AI Gateway key in Settings. Its base URL ends in /api/ai/v1: POST /chat/completions takes OpenAI's request, and GET /models lists the models you can name.
curl https://app.salexhub.com/api/ai/v1/chat/completions \
-H "Authorization: Bearer $SALEXHUB_AI_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "google/gemini-2.5-pro",
"messages": [
{ "role": "user", "content": "Summarise this week's late invoices." }
]
}'See who spent what. Nobody spends past their limit.
Every person, agent, workflow and API key has its own monthly limit, and the spend shows by who spent it. When one reaches its limit, that one stops, and everything else carries on.
How it is paid
On your own key
Calls to a provider whose key you added are billed by that provider, at your rates. Own keys come with Pro and Business.
On SalexHub
Everything else runs on SalexHub and is paid in AI credits. One AI credit is one US cent, and every plan shows what a task costs.
Run your business on the models you trust.
Start free on your own, and bring your keys when you are ready.