Skip to main content

​
Highlights

  • Reliable reverse-engineered channel: GPT image generation reverse-engineered model, behavior aligned with official capabilities
  • Highly competitive pricing: Flat $0.03/image, no resolution tiers, fully predictable cost
  • Three core capabilities: Text-to-image / single-image editing / multi-image fusion / natural-language editing — one-stop coverage
  • High text-rendering fidelity: Stable Chinese/English text rendering in signs, posters, and infographics
  • Native Chinese prompts: High-quality output without translation
  • 30-second generation: Returns in ~30s, R2 CDN accelerated downloads

​
Background

In April 2026, API易 officially launched gpt-image-2-all — a GPT image generation reverse-engineered model. While the official GPT-Image-1.5 uses tiered pricing (with high-quality images reaching $0.20 each), gpt-image-2-all offers a more economical choice at a flat $0.03/image per-call rate — ideal for scenarios that need “predictable cost + strong text rendering + Chinese friendliness”. For large-scale production of Chinese-language infographics, marketing posters, and social media content, this is a compelling option that preserves GPT-style image generation while pushing per-image cost to a new low.

​
Detailed Analysis

​
New Model

gpt-image-2-all

GPT image generation reverse-engineered modelFlat $0.03/image, ~30s generation, supports text-to-image, single-image editing, multi-image fusion, and natural-language editing. No need to worry about size/n/quality parameters — just describe size and style in the prompt.

​
Model Comparison

Featuregpt-image-2-allGPT-Image-1.5 (Official)Nano Banana 2
Channel typeReverse-engineeredOfficial directOfficial direct
BillingFlat per-callTiered by resolution + TokenPer-call / Token-based
Typical price$0.03/image$0.009 ~ $0.20/image$0.055/call
Generation speed~30s~10s~5-15s
Size controlVia promptsize parameteraspectRatio parameter
Chinese prompts✅ Native✅ Supported✅ Supported
Text rendering✅ High fidelity⭐⭐⭐⭐⭐ Best⭐⭐⭐⭐ Excellent
Multi-image fusion✅ image[] array✅ Edit endpoint✅ inlineData
Content restrictionsFewerStricterModerate
Choosing guide:
  • 💰 Cost-sensitive, Chinese-language scenarios → gpt-image-2-all (flat $0.03/image)
  • 🎨 Best text rendering, official direct → GPT-Image-1.5
  • 🍌 4K resolution, 14 aspect ratios → Nano Banana 2

​
Triple Endpoint Compatibility

gpt-image-2-all is compatible with three standard OpenAI-style endpoints:
  1. Text-to-image: POST /v1/images/generations (JSON)
  2. Image editing: POST /v1/images/edits (multipart/form-data)
  3. Chat-based: POST /v1/chat/completions (multi-turn with reference images)
You can reuse existing OpenAI SDKs and workflows seamlessly — just switch base_url and model.

​
Important Notes

Does not accept size/n/quality/aspect_ratio fieldsThis is an adaptive model; sending these fields may trigger parameter validation errors. Size and ratio must go directly into the prompt, and putting size words at the very front yields better adherence.Recommended phrasings:
  • Landscape 16:9 cinematic, old lighthouse at sunset
  • 1024×1024 square logo, minimalist cat line art
  • Portrait 9:16 phone poster, cyberpunk rainy night
b64_json field already contains the data URL prefixUnlike some legacy OpenAI-compatible models, this model’s b64_json field already includes the data:image/png;base64, prefix. You can use it directly as an HTML <img src>. Do not manually prepend the prefix, or you will produce a broken data URL.We recommend adding a startsWith('data:') compatibility check.

​
Practical Usage

​
Quick Start (Python)

import requests

response = requests.post(
    "https://api.apiyi.com/v1/images/generations",
    headers={"Authorization": "Bearer sk-your-api-key"},
    json={
        "model": "gpt-image-2-all",
        "prompt": "Landscape 16:9 cinematic, old lighthouse at sunset",
        "response_format": "url"
    },
    timeout=120
).json()

print(response["data"][0]["url"])

​
Multi-Image Fusion (cURL)

curl -X POST "https://api.apiyi.com/v1/images/edits" \
  -H "Authorization: Bearer sk-your-api-key" \
  -F "model=gpt-image-2-all" \
  -F "prompt=Put the person from image1 into the scene of image2, using the art style of image3" \
  -F "response_format=url" \
  -F "image[]=@./ref1.png" \
  -F "image[]=@./ref2.png" \
  -F "image[]=@./ref3.png"

​
Use Cases

Chinese Marketing Materials

Posters, livestream covers, social media graphics — stable Chinese text rendering

E-commerce Product Images

Multi-image fusion for product scene compositions — more flexible than compositing software

Infographics

Text-annotated infographics at a fraction of the cost of high-quality official models

Natural-Language Editing

No masks needed — iteratively edit images via natural-language descriptions

​
Pricing & Availability

ItemDetails
Model namegpt-image-2-all
Pricing$0.03 / image, per-call
Billing ruleFlat pricing — no tiers by resolution/quality/prompt length
Failed requestsNot charged (auth failures, parameter validation errors)
N images1 image per call; for N images call in parallel
PromotionStackable with deposit bonus activities
Ready to use: Existing API易 tokens can call this model directly — no special activation required.

​
Summary & Recommendations

The core value of gpt-image-2-all is “GPT style + Chinese friendly + predictable cost”:
  • ✅ If your business produces high-volume Chinese marketing content → recommended as your primary image generation channel
  • ✅ If you need the absolute best text rendering → use as a complement to GPT-Image-1.5 (cost-sensitive tasks on gpt-image-2-all, best-text tasks on 1.5)
  • ⚠️ If you need strict parameterized resolution/ratio control → Nano Banana 2 is a better fit (14 aspect ratios, 4K output)
Get started now:

Model Overview

Full capabilities and best practices

Text-to-Image Playground

Test text-to-image online

Image Editing Playground

Upload images and test editing/fusion

Deposit Bonus

See current deposit bonus promotions