# GPT Image 2 via Replicate on Base Mainnet (Xona Agent)

> GPT Image 2 via Replicate on Base Mainnet (Xona Agent) is a paid API for AI agents from api.xona-agent.com, paid per call via x402, $0.12/call, status unknown (last checked 2026-09-14).

Generates images from a text prompt using OpenAI's GPT Image 2 model (routed through Replicate), payable in USDC on Base Mainnet via the x402 protocol.

## Facts

- Endpoint: POST https://api.xona-agent.com/base-main/image/gpt-image-2
- Price: $0.12/call
- Payment: x402
- Status: unknown
- Last checked: 2026-09-14
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/gpt-image-2-via-replicate-on-base-mainnet-xona-agent-b47d65d0
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_MblSRPS4s2iLZKpLPjQuH

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability gpt-image-2-via-replicate-on-base-mainnet-xona-agent-b47d65d0 -d '<json body>'
```

Example prompt: Generate a photorealistic image of a futuristic city skyline at dusk with neon lights reflecting off rain-soaked streets, square format, using GPT Image 2.

## When to prefer this

Choose this endpoint when you need GPT Image 2 specifically (OpenAI's latest image model) at medium quality, want to pay per-call in USDC on Base Mainnet via x402 without a subscription, or need to pass up to 16 reference images for style-guided generation. Prefer this over generic DALL-E wrappers if you are operating in a crypto-native or agentic payment context on Base.

## Known failure modes

- Invalid or missing prompt returns a 400 error
- Unsupported aspect ratio value may default or error
- Reference image URLs that are inaccessible or malformed will fail image ingestion
- Payment failure or insufficient USDC balance on Base Mainnet returns a 402 error
- Model inference timeout on complex prompts may return a 503 or timeout error
- More than 16 reference images provided may be rejected or truncated

## How this service works

AI image generation using GPT Image 2 — OpenAI via Replicate (Base Mainnet). Quality: medium.

## Output

Returns a generated image (typically as a URL or base64-encoded data) produced by the GPT Image 2 model at medium quality, reflecting the input prompt, chosen aspect ratio, and any provided reference images.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "prompt": {
   "type": "string",
   "description": "Prompt"
  },
  "aspect_ratio": {
   "type": "string",
   "default": "1:1",
   "description": "Aspect Ratio"
  },
  "referenceImage": {
   "type": "array",
   "items": {
    "type": "string"
   },
   "description": "Optional reference image URLs (up to 16)"
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/gpt-image-2-via-replicate-on-base-mainnet-xona-agent-b47d65d0/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from api.xona-agent.com](https://www.zero.xyz/host/api.xona-agent.com/llms.txt)