Nataris is in beta and we're making it faster every day. Thanks for being here early.

Nataris

The People's AI Network

Unfiltered access to open-weight models — no content filtering, no model training, no vendor lock-in. Developers call an OpenAI-compatible API; the jobs route to Android phones worldwide. Phone owners keep 80%.

Developer

Sends AI request

Routing

Finds best match

Provider

Runs inference

For Developers

•No content filtering — open-weight models, your rules
•True data independence — your prompts run on community hardware, not inside a corporate AI data pipeline
•OpenAI-compatible — change two lines of code and go
•Growing P2P network — requests routed to real devices with automatic retry and fast failover
•$5 free credit — no credit card, no commitment

For Providers

•Earn from idle compute — from your phone, while you sleep or work
•80% is yours — withdrawn to UPI in India, payout link abroad
•No expertise needed — install from Google Play and tap Go Online
•Device protection — auto-pauses when hot or battery low

Two ways to use Nataris. Pick yours.

Build with Nataris Earn with your phone

Inference Capabilities

Text generation, multi-step workflows, and conversation memory — all through one OpenAI-compatible API, routed to the best available device on the network.

Text Generation with Llama 3.2

# OpenAI-compatible Chat Completions API
curl -X POST https://api.nataris.ai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nataris-balanced",
    "messages": [
      {"role": "user", "content": "Explain quantum computing in one paragraph."}
    ],
    "max_tokens": 256
  }'

3 chat models available

Qwen 2.5 0.5B

Text

Fast & reliable

Llama 3.2 1B

Text

General purpose

Phi-3 Mini 3.8B

Text

Code & reasoning

Try in Playground →

●No content filtering●No model training — your data never improves anyone's AI●80% is yours●$5 free credit

Already using OpenAI? One line to switch.

Nataris is OpenAI-compatible. Change the base URL — the rest of your code stays identical.

python

from openai import OpenAI

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

response = client.chat.completions.create(
    model="nataris-balanced"  # or nataris-fast / nataris-quality,
    messages=[{"role": "user", "content": "Hello"}]
)
print(response.choices[0].message.content)

Works with any OpenAI-compatible client. View full API docs →

What You Can Build

Open-weight models, your rules. Build what's hard to do on locked-down clouds.

Creative Freedom

Open-weight models without vendor-mandated content policies. No filtering on outputs — fiction, dialogue, experimental narratives, whatever you need to build.

Games, interactive fiction, creative writing tools, brainstorming apps, unrestricted chatbots.

Bots & Automation

Inference for community bots, game systems, and workflows.

Discord/Telegram bots, dynamic NPC dialogue, workflow automation, customer support.

Prototyping & Experimentation

$5 free credit, no credit card, multiple models to try. Good for MVPs, demos, and testing before committing to a vendor.

Early-stage products, model comparison, hackathons and learning.

Multi-Step AI Workflows

Research, code generation, and analysis — orchestrated across multiple inference steps with built-in budget controls.

Deep research, code review pipelines, document Q&A, agent-style reasoning.

Runs on People's Phones

Inference runs on independent phones, not in a data centre. We store the prompt and response with the job record for billing, support and disputes. They are never used to train models and never sold.

Why Nataris

No filtering — open-weight models, your rules

No model training — your data never improves anyone's AI

80% is yours — not AWS, not GCP, real providers

Resilient routing — automatic retry across multiple devices with fast failover

What Makes This Different

Nataris isn't another GPU cloud. It's an AI inference marketplace that runs on real devices — phones, not racks.

Mobile P2P Compute

io.net, Akash, Render — all datacenter GPUs. We route inference to smartphones. Billions of idle devices, zero capital cost to the network.

No Content Filtering

Open-weight models without vendor-mandated moderation. No content policies on outputs. You choose what you build.

80% is yours

Every dollar flows to real people, not data centers. Payouts to UPI in India; a secure payout link abroad.

Smart Failover

If a phone fails or drops offline, the job is usually sent to another phone.

Full Stack, Built

Text generation, multi-step workflows, conversation memory — live in beta.

Device Integrity Verification

Optional Google Play Integrity attestation for providers. Route to verified devices only.

●3 chat models + orchestration (growing)●Text, workflows, conversation memory●OpenAI-compatible API●Pay per token, no subscriptions

Be Among the First

The network is launching. Once providers start running inference and developers start sending requests, live stats will appear here — total inferences, tokens processed, and earnings paid to providers. All real numbers, nothing made up.

13+

Models Ready

Text

Chat models

80%

Is yours

Get API Access Become a Provider

For Developers

OpenAI-compatible API. Start in minutes.

Multi-step workflows and conversation memory — all built in.

$5 free API credits on signup — no credit card required

# Install and use like OpenAI
pip install openai

from openai import OpenAI

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

response = client.chat.completions.create(
    model="nataris-balanced"  # or nataris-fast / nataris-quality,
    messages=[{"role": "user", "content": "Hello!"}]
)

Get API Access View Documentation