Skip to main content
GPT-Image-1.5 is OpenAI’s latest flagship image generation model, released in December 2025. It surpasses its predecessor in speed, text rendering, and editing precision. Ranked #1 on LMArena Text-to-Image leaderboard with 1277 Elo.

​
Key Features

4x Speed Boost

Most images generate in under 10 seconds, 4x faster than GPT-Image-1

Superior Text Rendering

Supports dense text, markdown tables, and small typography for infographics and posters

Precision Editing

Preserves composition, lighting, and brand elements across 5+ editing iterations

Leaderboard #1

LMArena Text-to-Image 1277 Elo, Design Arena 1344, AA Arena 1272

​
Comparison with GPT-Image-1

FeatureGPT-Image-1.5GPT-Image-1
Generation SpeedUnder 10s (4x faster)Standard
Text RenderingExcellent (dense text, tables, small type)Good
Editing PrecisionHigh-fidelity element preservationStandard
Instruction FollowingMore precise and consistentStandard
Cost~20% cheaper than GPT-Image-1Standard pricing
LMArena Elo1277 (#1)-

​
Pricing

​
Token-Based Pricing

Token TypePrice
Input Tokens$5.00 / million tokens
Output Tokens$10.00 / million tokens

​
Per-Image Pricing

Low Quality
SizePrice
1024×1024$0.009 / image
1024×1536$0.013 / image
1536×1024$0.013 / image
Medium Quality
SizePrice
1024×1024$0.034 / image
1024×1536$0.051 / image
1536×1024$0.050 / image
High Quality
SizePrice
1024×1024$0.133 / image
1024×1536$0.200 / image
1536×1024$0.200 / image
The system automatically selects the more favorable billing method. Final charges are based on actual consumption.

​
API Usage

​
Endpoint

https://api.apiyi.com/v1/images/generations

​
Request Examples

curl https://api.apiyi.com/v1/images/generations \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "gpt-image-1.5",
    "prompt": "A professional infographic about AI trends in 2026, with charts, icons, and readable text labels",
    "size": "1024x1536",
    "quality": "high",
    "output_format": "png"
  }'

​
Request Parameters

ParameterTypeRequiredDescription
modelstringYesModel name, use gpt-image-1.5
promptstringYesImage description (max 1000 characters)
nintegerNoNumber of images, default 1, max 10
sizestringNo1024x1024, 1536x1024, 1024x1536, auto (default)
qualitystringNolow, medium, high, auto (default)
output_formatstringNopng (default), jpeg, webp
output_compressionintegerNoCompression level (0-100%), JPEG/WebP only
backgroundstringNotransparent, opaque, auto (default)
input_fidelitystringNoSet to high for high-fidelity preservation of first 5 input images during editing
response_formatstringNourl (default) or b64_json

​
Response Format

{
  "created": 1702486395,
  "data": [
    {
      "url": "https://..."
    }
  ]
}

​
Best Use Cases

​
Infographics and Typography

GPT-Image-1.5 has the strongest text rendering capability among AI image generation models:
  • Infographic generation
  • Product posters and marketing materials
  • Social media visual content
  • Data visualization with tables
  • Brand assets (precise logo and brand element preservation)
{/* Infographic example */}
response = client.images.generate(
    model="gpt-image-1.5",
    prompt="Create a clean infographic comparing 3 AI models: GPT-5, Claude 4, Gemini 3. Include performance bars, pricing info, and feature comparison table. Modern flat design, blue and white color scheme.",
    size="1024x1536",
    quality="high"
)

​
Product Photography and E-commerce

{/* Product showcase example */}
response = client.images.generate(
    model="gpt-image-1.5",
    prompt="Professional product photography of modern wireless earbuds case on white marble surface, soft studio lighting, 45-degree angle, clean background",
    size="1024x1024",
    quality="high"
)

​
Iterative Editing

GPT-Image-1.5 supports multi-round precise editing, maintaining consistency in composition, lighting, and brand elements across 5+ iterations.

​
FAQ

Per-image pricing is slightly higher (due to higher token unit prices), but it’s 4x faster with better quality. Token pricing is $5.00/M input and $10.00/M output. Actual cost per image varies based on prompt complexity.
Yes, simply change the model parameter from gpt-image-1 to gpt-image-1.5. The API format and parameters are fully compatible.
GPT-Image-1.5 can accurately generate dense text, markdown tables, and small typography. It ranks #1 on both LMArena Text-to-Image and Design Arena leaderboards.
Yes. Set background: "transparent" and use png or webp format to get transparent background images.
Generated image URLs are typically valid for 24 hours. We recommend downloading and saving them promptly.

GPT-Image Series Overview

View the complete GPT-Image series comparison and selection guide

GPT-Image-1

Classic image generation model, ideal for general use
OpenAI official documentation: platform.openai.com/docs/models/gpt-image-1.5