gpt-image-2 Image Generation

OpenAI-compatible gpt-image-2 text-to-image API: drop in by switching base_url, multi-region endpoints, unified billing with your QCode key

gpt-image-2 Image Generation¶

QCode.cc exposes a fully OpenAI-compatible gpt-image-2 text-to-image API. gpt-image-2 is OpenAI's latest text-to-image model (released April 2026) and currently has the strongest in-image text rendering of any public model — it can reliably render English and Chinese characters inside generated images.

What QCode.cc adds:

• Drop-in for the OpenAI SDK — only the base_url changes; request and response shapes are 100% identical
• Multi-region access: HK / Japan / US / EU / Shenzhen direct — pick the entry closest to your network
• Single API key: reuse your existing QCode.cc cr_ key — same quota account as Claude Code, Codex, and Gemini CLI
• Unified usage view: image calls and chat calls are merged in your dashboard, plus a dedicated self-service usage page

Common use cases: poster generation, illustrations, product imagery, UI mockups, social-media assets.

Quick start¶

Three lines of Python for your first image:

from openai import OpenAI
import base64

client = OpenAI(
    base_url="https://api.qcode.cc/qcode-img/v1",
    api_key="cr_YOUR_QCODE_API_KEY",
    timeout=180.0,  # single image takes 30-90s; set at least 180s
)

result = client.images.generate(
    model="gpt-image-2",
    prompt="A cyberpunk Tokyo street at night, neon reflecting in rain puddles",
    size="1024x1024",
    quality="low",
    n=1,
)

with open("output.png", "wb") as f:
    f.write(base64.b64decode(result.data[0].b64_json))

⚠️ Always set timeout ≥ 180s explicitly: the OpenAI SDK's default timeout is too short. gpt-image-2 is a reasoning-based model and takes much longer than traditional text-to-image (see Generation latency).

Endpoints¶

gpt-image-2 is available on every QCode.cc access domain — append /qcode-img/v1 to the host:

All entries route to the same billing system; usage and quota are unified. For the full domain reference, see Endpoints and API Paths.

qcode-img is the path prefix dedicated to image generation, parallel to /api (Anthropic), /openai/v1 (OpenAI), and /gemini (Google) used by other protocols.

API Reference¶

Endpoint¶

POST {base_url}/images/generations

Headers¶

Request body¶

{
  "model": "gpt-image-2",
  "prompt": "A small ceramic vase with sunflower, photorealistic",
  "size": "1024x1024",
  "quality": "low",
  "n": 1
}

Response¶

{
  "created": 1777135432,
  "data": [
    {
      "b64_json": "iVBORw0KGgo...(base64 PNG, 1-3 MB)",
      "revised_prompt": "A small ceramic vase with sunflower..."
    }
  ]
}

• b64_json: base64-encoded PNG. Render directly with <img src="data:image/png;base64,...">
• revised_prompt: model's polished version of your prompt (optional to display)

Error response¶

Errors follow the standard OpenAI error schema:

{
  "error": {
    "type": "rate_limit_error",
    "code": "image_daily_limit",
    "message": "Daily image generation count limit reached..."
  }
}

Code samples¶

Python (OpenAI SDK, recommended)¶

from openai import OpenAI
import base64

client = OpenAI(
    base_url="https://api.qcode.cc/qcode-img/v1",
    api_key="cr_YOUR_QCODE_API_KEY",
    timeout=180.0,
)

result = client.images.generate(
    model="gpt-image-2",
    prompt="A cyberpunk Tokyo street at night, neon reflecting in rain puddles",
    size="1024x1024",
    quality="low",
    n=1,
)

img_bytes = base64.b64decode(result.data[0].b64_json)
with open("output.png", "wb") as f:
    f.write(img_bytes)
print("Saved output.png")

curl¶

curl https://api.qcode.cc/qcode-img/v1/images/generations \
  -H "Authorization: Bearer cr_YOUR_QCODE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-image-2",
    "prompt": "A cyberpunk Tokyo street at night",
    "size": "1024x1024",
    "quality": "low",
    "n": 1
  }' \
  | jq -r ".data[0].b64_json" | base64 -d > output.png

JavaScript / Node.js / browser¶

const r = await fetch("https://api.qcode.cc/qcode-img/v1/images/generations", {
  method: "POST",
  headers: {
    "Authorization": "Bearer cr_YOUR_QCODE_API_KEY",
    "Content-Type":  "application/json",
  },
  body: JSON.stringify({
    model: "gpt-image-2",
    prompt: "A cyberpunk Tokyo street at night",
    size: "1024x1024",
    quality: "low",
    n: 1,
  }),
});
const json = await r.json();
const dataUrl = "data:image/png;base64," + json.data[0].b64_json;
document.querySelector("img").src = dataUrl;

Limits and quotas¶

Default limits¶

Defaults are sufficient for the vast majority of users. Contact support to raise them.

Usage queries¶

• Customer dashboard: image calls are merged with chat calls (last-used time, daily cost, total cost, model breakdown)
• Self-service page: https://api.qcode.cc/qcode-img/usage — paste your API key to see 30-day usage, detailed call list, and ECharts trend graph (your API key is stored only in your browser, never uploaded)

Billing¶

Per-image pricing¶

$0.08 per-call floor¶

• If actual cost < $0.08, you are billed $0.08 (low / medium quality typically hit this floor)
• If actual cost ≥ $0.08, you are billed the actual amount (not raised)

Multiple images¶

With n > 1, cost scales linearly. Example: n=2 + 1024×1024 high = 2 × $0.211 = $0.422.

Failed requests not billed¶

Any 4xx / 5xx failure is not billed. Client-disconnect (connection closed mid-request) is also not billed.

Currency¶

Pricing is in USD; final settlement follows the currency policy of your main QCode.cc account (CNY / USD).

Generation latency and timeout¶

gpt-image-2 is a reasoning-based model — significantly slower than traditional text-to-image (DALL·E 3 / SDXL):

Practical guidance:

• The OpenAI Python SDK has a short default timeout — always set timeout=180.0 or higher
• Browser fetch has no default timeout, but if you use an AbortController, give it at least 180 s
• Mainland China users on medium / high should use the Shenzhen direct entry, otherwise the CDN 524 issue below will hit you

CDN 100s hard limit (524 errors)¶

HTTPS requests through api.qcode.cc / asia.qcode.cc / eu.qcode.cc / us.qcode.cc are fronted by a global CDN (CloudFlare). The CDN forces a 524 error if a single request waits more than 100 seconds for the origin to respond.

Workarounds (recommended for medium / high):

1. Use the Shenzhen HTTP entry http://103.236.53.153/qcode-img/v1 directly — no CDN, no 100s limit
2. Or accept occasional 524 and add client-side retry

Prompt tips¶

• Multi-language: write your prompt in English, Chinese, or mixed — all work
• Be specific: scene, composition, lighting, style, lens / focal length / camera angle, etc.
• Avoid brand names / public figures: the model may refuse or return blurry results (OpenAI content policy)
• Text rendering: gpt-image-2 excels at rendering text inside images — embed English / Chinese titles, short phrases, or poster text directly in the prompt; no special syntax required

Sample prompt:

A vintage poster in Bauhaus style, bold black text "MORNING COFFEE" centered,
warm orange and cream color palette, geometric shapes, slightly textured paper background

Differences from the OpenAI API¶

Online Playground¶

https://api.qcode.cc/qcode-img/ — try in the browser:

• Paste API key + prompt, generate immediately
• Bilingual UI (EN / ZH)
• Defaults to low quality (avoids CDN 524)
• Built-in API reference (curl / Python / JavaScript tabs, parameter table, error codes)
• One-click PNG download

Related¶

• Endpoints and API Paths — every QCode.cc access domain, protocol path, and curl smoke-test
• Billing — plans, quotas, cost rules