Model updates

GPT Image 2 is live

BetterToken updated GPT Image 2 on April 29, 2026 for high-quality image generation, posters, marketing assets, and visual drafts. Includes setup guidance.

Release date: April 29, 2026

BetterToken has added the gpt-image-2 image generation model entry. Check the model plaza for the currently available model ID, then call it through the OpenAI-compatible image generation API.

OpenAI introduced ChatGPT Images 2.0 on April 21, 2026 and lists gpt-image-2 as the latest GPT Image model in its developer docs. OpenAI describes GPT Image 2 as a model for fast, high-quality image generation and editing, with text and image input, image output, flexible sizes, and high-fidelity image inputs.

Highlights

Higher image quality: OpenAI marks GPT Image 2 as its highest-performance image generation model, suited for polished visual assets
More flexible sizes: gpt-image-2 supports more valid image dimensions, including common square, landscape, portrait, 2K, and 4K use cases; sizes above 2K are currently experimental
Higher-fidelity image input: In image editing and reference-image workflows, gpt-image-2 processes image inputs at high fidelity, useful for product photos, portraits, and brand assets
Stronger text and complex information: OpenAI’s system card says ChatGPT Images 2.0 improves world knowledge, instruction following, dense text, and complex detail generation
Streaming partial images: The Image API can return partial images through stream and partial_images, making product experiences feel more responsive

Good use cases

Product launch posters, campaign assets, and social images
Illustrations for docs, blogs, and long-form articles
Visual concept drafts for apps, websites, and SaaS interfaces
Product images, advertising images, and brand direction exploration
Visual content that includes Chinese or multilingual text

API setup

For the OpenAI-compatible protocol, use:

Setting	Value
`Base URL`	`https://bettertoken.ai/v1`
Authentication	`Authorization: Bearer YOUR_API_KEY`
Model ID	`gpt-image-2`
Endpoint	`POST /v1/images/generations`

curl https://bettertoken.ai/v1/images/generations \
  -H "Authorization: Bearer $BETTERTOKEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-image-2",
    "prompt": "Create a modern Chinese launch poster for an AI coding tool, with a clear headline, product UI, and clean background",
    "size": "1024x1024",
    "quality": "high"
  }' | jq -r '.data[0].b64_json' | base64 --decode > gpt-image-2.png

For the full parameter list, see the text-to-image documentation.

Setup guidance

If you use your own application, script, or external tool, confirm that it supports the OpenAI-compatible image generation endpoint. Image generation uses https://bettertoken.ai/v1; do not use the Claude Code / Anthropic protocol endpoint.

Copy the current model ID from the model plaza. Do not use ChatGPT UI mode names as API parameters.

Notes

Image generation averages about 100 to 200 seconds. Set client, gateway, or external-tool request timeouts above 300 seconds, and show a waiting state in your application.
gpt-image-2 currently does not support transparent backgrounds. Generate a non-transparent image first, then remove the background with another image tool if needed.
Image results are returned as b64_json by default. Decode the base64 value before saving the image.
Large, high-quality images can increase latency and cost. Use quality: "low" for previews.
Review images with text, layout, and brand elements before publishing them.