Nataris — The People's AI Network

AI infrastructure owned by those who power it.

Website • API Docs • Pricing • FAQ

---

What is Nataris?

Developers get affordable, private inference via API. Providers earn by running models on their mobiles—and keep 85%. The network grows with every device that joins.

Powered by people, not datacenters.

---

How It Works

Developer App  →  Nataris API  →  Provider Network  →  AI Response
                      ↓
              Smart Routing
         (model, latency, availability)
                      ↓
            On-Device Inference

1. Developers send API requests to Nataris
2. Smart routing finds the best available device
3. Providers run inference locally on their phones
4. Results are returned via secure connections

---

For Developers

Benefit	Details
$5 free credits	No credit card required
Pay for what you use	No vendor lock-in, no minimums
Open-weight models	You choose how you use them
No model training	Your prompts are never used to train AI models
OpenAI-compatible	Swap your base URL in minutes

Quick Start

curl -X POST https://api.nataris.ai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "llama-3.2-1b-instruct-q4_k_m",
    "messages": [{"role": "user", "content": "Explain quantum computing in one paragraph."}]
  }'

OpenAI SDK Compatible

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_NATARIS_KEY",
    base_url="https://api.nataris.ai/v1"
)

response = client.chat.completions.create(
    model="llama-3.2-1b-instruct-q4_k_m",
    messages=[{"role": "user", "content": "Hello!"}]
)

---

For Providers

Benefit	Details
Joining bonus	Earn a bonus after completing your first job (subject to eligibility criteria)
Earn from idle compute	Your phone works while you don't
No expertise needed	Just install and go online
Device protection	Thermal and battery safeguards
Keep 85%	Fair revenue share

Get Started

1. Download the Nataris app
2. Select models to host
3. Go online when you want to earn
4. Get paid per inference

Requirements: Android 8.0+, 4GB+ RAM

---

Inference Capabilities

Category	Models	Use Cases
Text Generation	Qwen 2.5 0.5B, Llama 3.2 1B, Phi-3 Mini 3.8B	Chat, code, summarization
Text Generation (coming soon)	Mistral 7B, Llama 2 7B	Advanced tasks, broad knowledge

Audio (STT, TTS, Voice Agent): Built but temporarily disabled while we scale the text inference network. Will be re-enabled once provider capacity grows.

Advanced Features

Feature	Description
Multi-Step Workflows	Orchestrate research, code gen, agent, and map-reduce pipelines via a single API call
RAG (Document Q&A)	Upload documents, get answers grounded in your content (coming soon)
Conversation Memory	Server-side message persistence with auto-summarization
Cost Controls	Budget caps per workflow, cost estimation endpoint

Orchestration Example

response = client.chat.completions.create(
    model="llama-3.2-1b-instruct-q4_k_m",
    messages=[{"role": "user", "content": "Research renewable energy trends"}],
    extra_body={
        "orchestration": {
            "enabled": True,
            "workflow": "research",
            "max_cost_usd": 1.0
        }
    }
)

---

What You Can Build

• Creative freedom — Open-weight models, no content filtering, your rules
• Privacy-first apps — Your prompts are not used to train models or fed into corporate AI pipelines
• Bots & automation — High volume without strict rate limits
• Prototyping — Test ideas without big upfront spend
• Research pipelines — Multi-step analysis orchestrated automatically
• Document Q&A — RAG-powered answers from your own documents (coming soon)

---

Why Nataris?

	Nataris	Traditional APIs
Data use	Prompts never used for model training	Often used to improve models
Models	Open-weight, unfiltered	Vendor-controlled
Pricing	Pay-per-use, no minimums	Subscriptions, limits
Economics	85% to providers	Value to corporations
Filtering	No content filtering	Vendor-controlled output

---

Open Source

Built on amazing open-source projects:

Project	Purpose
RunAnywhere	Cross-platform AI inference
llama.cpp	Efficient LLM inference
ONNX Runtime	Cross-platform ML
Ollama	Cloud backstop inference

---

Documentation

• API Reference — Complete endpoint docs (orchestration, conversations)
• Integration Guide — Step-by-step tutorial with advanced features
• Security — Our security model
• FAQ — Common questions

---

Code Examples

Check out the examples folder:

• Node.js — Express.js integration
• Python — FastAPI integration
• cURL — Command-line examples

---

Roadmap

Quarter	Milestone
Q1 2026	Beta launch — Android app, core models ← We are here
Q2 2026	iOS app, 7B models, RAG
Q3 2026	Enterprise tier, SLAs
Q4 2026	Ecosystem — Connectors, governance

---

Contact

• Website: nataris.ai
• Support: support@nataris.ai
• Security: security@nataris.ai

---

License

This repository (documentation and examples) is licensed under the MIT License. See LICENSE.

---

"AI infrastructure, owned by everyone."

Powered by people, not datacenters.