# Gab AI

> All the top AI models in one place. The only right-wing Christian AI, and the only platform with complete privacy.

## Platform Overview

Gab AI is an independent AI platform built outside the control of Big Tech. We provide access to multiple top-tier AI models through a unified interface while respecting your privacy and freedom. Our platform offers text generation, image generation, video generation, music generation, and text-to-speech capabilities.

### Privacy Protection

Gab AI protects user privacy by masking identities from third-party AI providers:
- When using models like GPT or Claude through Gab AI, the provider sees the prompt but NOT who sent it
- User identities are masked from all third-party AI providers
- Generalized batch inference requests instead of personal identity tracking
- No data harvesting or profiling
- Your conversations stay private
- No use of personal information for marketing
- No data sales to advertisers
- No targeted advertising or tracking across the web

### Values & Philosophy

- Uncensored and unfiltered AI platform
- Christian values through our proprietary Arya AI model
- Committed to free speech and First Amendment principles
- Independent from Big Tech control and influence
- No political censorship or bias
- Augmented Intelligence to enhance human capabilities, not replace them

## Company Information

**Name:** GAB AI INC
**Location:** Clarks Summit, Pennsylvania, United States
**Founded:** 2016
**Type:** Private Company

**Address:**
Gab AI Inc.
1143 Northern Blvd #158
Clarks Summit, PA 18411
United States of America

**Contact:**
- Support: ai@gab.com
- Legal: legal@gab.ai
- General: support@gab.ai

## Available AI Models

### Text Models

Gab AI provides access to all major AI language models:

**Gab Proprietary:**
- Arya - Gab's primary AI model, designed for broad applications with Christian values and conservative principles. Maintains high impartiality and neutrality.

**OpenAI:**
- GPT-5 - Latest flagship model
- GPT-5.1 - Enhanced version
- GPT-4o - Multimodal model
- GPT-4.1 - Previous generation
- o3 - Reasoning model
- o3-mini - Compact reasoning model
- o4-mini - Latest compact reasoning model

**Anthropic:**
- Claude Opus 4.7 - Most capable Claude model with step-change improvement in agentic coding
- Claude Opus 4.6 - Premium reasoning and analysis
- Claude Opus 4.5 - Previous generation flagship
- Claude Sonnet 5 - Latest Sonnet
- Claude Sonnet 4.6 - Balanced performance
- Claude Haiku 3.5 - Compact model

**Google:**
- Gemini 3 Pro - Latest Gemini
- Gemini 2.5 Pro - Advanced reasoning
- Gemini 2.5 Flash - Fast responses
- Gemini 2.0 Flash - Previous generation

**DeepSeek:**
- DeepSeek V3 - Advanced open model
- DeepSeek R1 - Reasoning model

**Alibaba:**
- Qwen 3 235B - Large language model
- Qwen 3 235B Thinking - Reasoning variant

**Moonshot:**
- Kimi K3 - Frontier multimodal reasoning
- Kimi K2.6 - Next-gen multimodal model for long-horizon coding and agent orchestration

**Other:**
- Hermes 4 405B (NousResearch)

### Image Generation Models

- GPT Image 2 - OpenAI state-of-the-art image generation and editing
- GPT Image 1 - OpenAI high quality
- GPT Image 1 Mini - OpenAI fast generation
- Imagen 4 - Google latest
- Seedream 4 - ByteDance model
- Qwen Image - Alibaba image model
- Gemini 2.5 Flash Image - Google multimodal

### Video Generation Models

- Veo 3.1 - Google high quality video
- Veo 3.1 Fast - Google fast video generation
- Sora 2 - OpenAI video generation
- Kling 2.5 Turbo Pro - Advanced video
- Kling 2.5 Turbo Pro (Image-to-Video) - Animate images
- Hailuo 2.3 - MiniMax video
- Hailuo 2.3 Pro - MiniMax premium
- Wan 2.5 - Alibaba video

### Music Generation Models

- MiniMax Music 2.0 - Create songs and music tracks

### Text-to-Speech Models

- GPT-4o Mini TTS - Natural voice synthesis with multiple voice options

## Pricing Plans

### Free Plan (with confirmed email)

- 30 messages per 3 hours
- 5 images per day
- 10 visible conversations
- 10 shortcuts
- 1 custom agent
- Images expire in 3 hours
- Access to Arya model
- Credit-based system for additional usage

### Plus Plan - $20/month or $150/year

- 300 messages per hour (in-app)
- 300 images per day
- Unlimited visible conversations
- Unlimited shortcuts
- Unlimited custom agents
- Free default model messages in app (not included in API)
- Images expire in 7 days
- Can regenerate/re-roll responses
- 10,000 API requests per day (uses credits)
- Access to ALL models (GPT-5, Claude, Gemini, etc.)
- 2,000 monthly credits included
- Advanced agent settings
- Custom image dimensions
- Pin top conversations
- Early access to new features

## Credit System

Gab AI uses a credit-based system for usage beyond plan limits.

### How Credits Work

- **Monthly Allotment:** Plus subscribers receive 2,000 credits per month (resets on the 1st)
- **Purchased Credits:** Never expire, used after monthly credits are exhausted
- **Default Model:** Free for Plus subscribers in the app (not included in API)
- **API Arya:** 1 credit per 32,000 input tokens (minimum 1 per request). In-app Arya for Plus is unchanged and still free. Paid models use their listed credit costs, which already scale with context.

### Credit Costs by Action

**Text Models:**
- Base cost: 1-10 credits depending on model
- Cost increases with longer conversations (context size)
- Context threshold typically 20,000 tokens per credit increment

**Image Generation:**
- 5-20 credits depending on model and settings

**Video Generation:**
- 20-100 credits depending on model and duration

**Audio/TTS:**
- 1-5 credits based on text length

**Web Search:**
- +1 credit for web search requests

**Deep Research:**
- +2 credits for comprehensive research

**File Uploads:**
- 1 file: +1 credit
- 5+ files: +2 credits
- 9+ files: +3 credits

### Purchasing Credits

Additional credits can be purchased at any time and never expire. Visit the billing settings to purchase credit packs.

## Intent Detection Engine

Gab AI includes an intelligent intent detection system that automatically understands what you want to do.

### Supported Intents

- **answer** - General questions and conversations
- **search-web** - Real-time web search for current information
- **research** - Deep research requiring comprehensive analysis
- **analyze-url** - Analyze content from URLs you share
- **create-image** - Generate images from text descriptions
- **edit-image** - Transform or style existing images (image-to-image)
- **create-video** - Generate videos from text
- **image-to-video** - Animate images into videos
- **create-music** - Generate songs and music
- **use-reasoning** - Complex tasks requiring step-by-step reasoning

### Automatic Model Routing

When you don't select a specific model, Arya (our intelligent router) automatically:
1. Detects your intent from your message
2. Selects the best model for the task
3. Routes to image/video/music models when creative content is requested
4. Enables web search when current information is needed
5. Uses reasoning models for complex problem-solving

## Core Features

### Web Search Integration

- Real-time web search across AI models
- Up-to-date information access
- Search results integrated directly with AI responses
- Automatic detection of queries needing current information

### Deep Research Mode

- Comprehensive multi-source research
- Detailed analysis and synthesis
- Academic-level research capabilities
- Automatic activation for complex research queries

### File Management

- Upload and analyze documents with AI
- Chat with your files (PDFs, code, documents)
- Document understanding and extraction
- Image analysis and description
- Support for multiple file types

### Collections

- Organize conversations into collections
- Custom instructions per collection
- Shared file context across every chat in a collection
- Per-collection default model and settings for new chats

### Custom AI Agents

- Build personalized AI assistants
- Configure custom system prompts
- Set agent personality and behavior
- Create image generation agents
- Save and reuse custom agents

### Memory System

- AI remembers important information about you
- Personalized responses based on context
- Memory suggestions for key information
- Privacy-focused memory management

### Voice Mode

- Talk to AI using voice input
- Natural conversation flow
- Text-to-speech responses available

## API Documentation

### Overview

Gab AI provides an OpenAI-compatible API for developers. Access all models through a single, unified endpoint.

### Base URL

```
https://gab.ai/v1
```

### Authentication

Include your API key in the Authorization header:

```
Authorization: Bearer YOUR_API_KEY
```

Get your API key from Settings > API Settings in your dashboard.

### Rate Limits

- Plus subscribers: 10,000 requests per day
- Rate limit headers included in all responses:
  - X-RateLimit-Limit
  - X-RateLimit-Remaining
  - X-RateLimit-Reset

### API Endpoints

**Chat Completions:**
```
POST /v1/chat/completions
```
Generate text completions. Fully OpenAI SDK compatible. Supports streaming (`stream: true`), function calling (`tools` + `tool_choice`), structured outputs (`response_format` with `text`, `json_object`, or strict `json_schema` modes), and live web search (`web_search: true`) on web-search-capable models.

**Image Generation:**
```
POST /v1/images/generations
```
Generate images from text prompts.

**Image Editing (reference / input images):**
```
POST /v1/images/edit
POST /v1/images/edits   (OpenAI-compatible alias)
```
Edit one or more existing images based on a text prompt — this is how you pass a reference/input image to models like `flux-2-pro` (do not put `image` on `/v1/images/generations`, and do not use `/v1/chat/completions` for image-only models). Accepts source images two ways: multipart/form-data with repeated `image` fields (up to 10), or a JSON body with `image` as a file id (or array of ids) previously uploaded via `POST /v1/files` with `purpose: "vision"`. Pass a `prompt` describing the change. The API equivalent of the in-app AI Image Editor — useful for character-consistent generation, compositing two images into one, restyling a photo, removing/replacing objects, or swapping backgrounds. Defaults to `gpt-image-1-mini`; supports `n` (1-4), `size`, `quality`, and `async: true` (poll `GET /v1/images/edits/:jobId`). Returns OpenAI-style `{ created, data: [{ url, revised_prompt }], usage }`.

**Video Generation:**
```
POST /v1/videos/generations
GET /v1/videos/:jobId
```
Submit video generation (async) and poll for status.

**Text-to-Speech:**
```
POST /v1/audio/speech
```
Convert text to natural speech.

**Embeddings:**
```
POST /v1/embeddings
```
Generate vector embeddings for text inputs. Useful for semantic search, RAG, and clustering.

**List Models:**
```
GET /v1/models
GET /v1/models/{id}
```
List all available models. Authentication is optional. Each model includes `aliases` (gpt-5.5, anthropic/claude-sonnet-4-5), `context_window`, `max_output_tokens`, `capabilities.function_calling`, and `recommended_for` (coding, agents, image, …).

**Agent setup (Cursor, OpenCode, OpenClaw, Claude Code):**
```
GET /v1/agent-setup?client=cursor
GET /openapi.json
GET /llms-full.txt
```
Generated configs for coding agents. OpenAPI 3.1 spec is at /openapi.json.

**Responses API:**
```
POST /v1/responses
GET /v1/responses/{id}
```
OpenAI Responses API with function tools, streaming argument deltas, and `previous_response_id` for multi-turn agent loops.

**Anthropic Messages:**
```
POST /v1/messages
POST /v1/messages/count_tokens
```
Claude Code compatible. Honors `cache_control` on system and tools. Agentic requests auto-mark the last tool as a cache breakpoint when the caller does not set one. Usage includes `cache_read_input_tokens`. Cache-hit prompt tokens are billed at a 90% credit discount (minimum 1 credit).

**Chat Completions (coding agents):**
```
POST /v1/chat/completions
```
Supports streaming tool-call argument deltas, `reasoning_effort`, `reasoning_content` in the stream, and `prompt_tokens_details.cached_tokens` in usage. Model ids accept aliases (`gpt-5.5`, `openai/gpt-5-5`, `anthropic/claude-sonnet-4-5`, date-stamped Claude ids).

**Check Credits:**
```
GET /v1/credits
```
Check your credit balance.

**Usage History:**
```
GET /v1/usage
```
Per-request usage records with tx_id, model, tokens, context_tokens, credits_used, and response time. Supports `after`/`before` tx_id cursors, `start_date`/`end_date` (ISO or unix), plus `endpoint` and `model` filters for programmatic billing audits.

### Example: Python

```python
from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_KEY",
    base_url="https://gab.ai/v1"
)

response = client.chat.completions.create(
    model="arya",
    messages=[
        {"role": "user", "content": "Hello!"}
    ]
)
print(response.choices[0].message.content)
```

### Example: Node.js

```javascript
import OpenAI from 'openai';

const client = new OpenAI({
    apiKey: 'YOUR_API_KEY',
    baseURL: 'https://gab.ai/v1'
});

const response = await client.chat.completions.create({
    model: 'arya',
    messages: [{ role: 'user', content: 'Hello!' }]
});
console.log(response.choices[0].message.content);
```

### Example: cURL

```bash
curl https://gab.ai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"arya","messages":[{"role":"user","content":"Hello!"}]}'
```

### Structured Outputs (response_format)

Pass an OpenAI-compatible `response_format` on `/v1/chat/completions` to constrain the model's output:

- `{"type": "text"}` — plain text (default).
- `{"type": "json_object"}` — free-form JSON.
- `{"type": "json_schema", "json_schema": {"name": ..., "strict": true, "schema": {...}}}` — strict JSON Schema enforcement. The assistant's `message.content` is a JSON-encoded string that conforms to the supplied schema.

Anthropic (Claude) models do not support `response_format` natively — for structured output on Claude, define a function whose `parameters` is your JSON Schema and pass it via `tools`. See the function-calling docs.

### Web Search

The platform can perform a live web search for a request and inject the results into the model's context (the model then cites its sources). This runs only on models that advertise `capabilities.web_search: true` (for example `arya`) and only when you opt in:

- `web_search: true` — explicit toggle (recommended for questions that need fresh information).
- `web_search_options: {}` — OpenAI-compatible alternative; providing the object (even empty) enables search.
- A plain-language request such as "search the web for ..." or "google it" also triggers a search.

For other models, or when none of the above are present, no search is performed and the model answers from its training data. Available on `/v1/chat/completions`, `/v1/responses`, and `/v1/messages`.

### API Credit Costs

Credit costs vary by model, context length, and generation type. Models that are free in-app (Arya) cost 1 credit per 32,000 estimated input tokens on the API, with a minimum of 1 credit per request. Prompt-cache hits are billed at a 90% credit discount (still minimum 1 credit). Paid models use their listed per-request costs, which already scale with context.

## Subscription & Billing

### Payment Methods

- Credit/Debit cards (Visa, Mastercard, American Express, Discover)
- Bitcoin payments available

### Subscription Management

- Cancel anytime from Settings > Plan & Billing
- Access continues until end of billing period after cancellation
- Change payment method in billing settings

### Refund Policy

All subscriptions and credit purchases are final and non-refundable due to upfront computational resource allocation. Limited exceptions may apply for billing errors. See full Refund Policy at /refund-policy

## Legal & Policies

### Terms of Service

- Users must be 18 years or older
- Platform governed by Delaware law
- Binding arbitration for disputes
- See full Terms of Service at /terms-of-service

### Privacy Policy

- Comprehensive privacy protections
- Identity masking from third-party providers
- Data processed in the United States
- See full Privacy Policy at /privacy-policy

### Terms of Sale

- Automatic renewal for subscriptions
- Cancellation available anytime (effective at end of billing period)
- See full Terms of Sale at /terms-of-sale

## FAQ

### What is Arya?

Arya is Gab's primary AI model, designed for broad applications and trained to maintain a high level of impartiality and neutrality with Christian values. When you use Gab AI without selecting a specific model, Arya intelligently routes your request to the best model for the task.

### Do default model messages cost credits?

In the Gab AI app, Plus subscribers can chat with the default Arya model for free. Not included in API — Arya API requests cost 1 credit per 32,000 input tokens (minimum 1 per request).

### Do I need Plus to use Gab AI?

No, Gab AI offers a free tier with limited usage. The free plan includes 30 messages per 3 hours and 5 images per day. Plus subscribers get significantly higher limits, access to all premium models, and additional features.

### How does privacy protection work?

When you use third-party models (GPT, Claude, Gemini, etc.) through Gab AI, we act as an intermediary. The AI provider receives your prompt but has no way to identify you personally. Your identity is masked at the infrastructure level.

### Can I use Gab AI for commercial purposes?

Yes, you can use Gab AI for commercial purposes. API access is available for Plus subscribers to build applications and integrations.

### How do I contact support?

Email us at ai@gab.com for any questions or issues.

## Related Links

- Main Platform: https://gab.ai
- Social Network: https://gab.com
- News & Blog: https://news.gab.com
- YouTube: https://www.youtube.com/@GabdotAI
- X (Twitter): https://x.com/Gab__AI

## Last Updated

January 2026