Now running DeepSeek R1, Llama 3.1 & more

Frontier AI. In Your Control.

Self-hosted AI platform powered by open-source models. Web search, memory, skills, full API access — your data never leaves your server.

Start for free → Read the docs

Quick start — 30 seconds

# OpenAI-compatible API
curl https://bundleai.cloud/v1/chat/completions \
  -H "Authorization: Bearer YOUR_KEY" \
  -d '{"model":"deepseek-r1:14b",
      "messages":[{"role":"user","content":"Hello!"}]}'

# Response streams in real-time
→ "Hello! How can I assist you today?"

Platform Features

Everything you need.
Nothing you don't.

BundleAI packs enterprise-grade AI capabilities into a clean, fast platform anyone can run.

🧠

Persistent Memory

AI remembers your conversations across sessions. Pin facts, preferences, and context that always stay with you.

🔍

Live Web Search

Automatic real-time web search for current events, prices, news, and anything beyond training data.

⚡

Smart Routing

Messages auto-route to the right model — fast 3B for greetings, powerful 14B for complex reasoning.

📚

RAG Knowledge Base

Upload PDFs, docs, and text files. AI answers from your documents with full source attribution.

🎯

Custom Skills

Create specialized AI personas — code reviewer, researcher, writer, analyst — with custom system prompts.

🔌

OpenAI-Compatible API

Drop-in replacement for OpenAI. Same API format — switch any app by changing one URL.

🧮

Built-in Tools

Calculator, weather, URL fetcher, web search, page summarizer — all available to the AI automatically.

🔐

Full Data Privacy

Every conversation stays on your server. No third-party APIs. No data logging. 100% self-hosted.

💳

Token Economy

Pay per use with Stripe or crypto. Top up tokens, track usage, manage users from the admin panel.

Intelligent Chat

Thinks before it answers.

DeepSeek R1's reasoning model shows its thought process — collapsible thinking blocks let you see exactly how it arrived at every answer.

✓ Collapsible "💭 Thinking..." reasoning blocks
✓ Real-time streaming responses
✓ Multi-turn conversation memory
✓ Code syntax highlighting with copy buttons
✓ 5 themes including dark, midnight, forest

👤

Explain how neural networks learn

🧠

💭 Reasoning · 2.1s

Let me think through backpropagation and gradient descent...

Neural networks learn through **backpropagation** — adjusting millions of weights to minimize prediction error. Each training example nudges the network slightly closer to the right answer...

deepseek-r1:14b · streaming

Developer API

Switch from OpenAI in 30 seconds.

BundleAI speaks OpenAI's API format natively. Change one URL and your existing apps just work — with full streaming, tool calls, and vision support.

✓ OpenAI-compatible /v1/chat/completions
✓ Personal API key management
✓ Token-based rate limiting
✓ Webhook support for integrations
✓ Full API documentation at /docs

Python example

from openai import OpenAI

client = OpenAI(
  base_url="https://bundleai.cloud/v1",
  api_key="bai_your_key_here"
)

response = client.chat.completions.create(
  model="deepseek-r1:14b",
  messages=[{
    "role": "user",
    "content": "Hello!"
  }]
)

print(response.choices[0].message.content)

How it works

Up and running in minutes.

Create account

Choose your model

Select from 7+ models — DeepSeek for reasoning, Llama for speed, or pull any Ollama model.

Start chatting or building

Use the chat UI directly or integrate the API into your app with one URL change.

Top up when needed

Buy token packs with Stripe or crypto. Tokens never expire. Pay only for what you use.

Pricing

Pay for what you use.

Tokens never expire. Start free with 100K tokens. Scale as you grow.

Starter

$4.99

500K tokens

Get started

Basic

$12.99

2M tokens