OpenAI Releases ChatGPT Images 2.0: Smarter, Multi-Image Generation

OpenAI has unveiled ChatGPT Images 2.0, an updated image generation model now available to all ChatGPT and Codex users. The update introduces powerful new capabilities, including multi-image generation from a single prompt, improved text rendering across multiple languages, and access to advanced reasoning features for paying subscribers.

"Images 2.0 brings an unprecedented level of specificity and fidelity to image creation." — OpenAI press release

Availability and Pricing

The model launched Tuesday and is accessible across three tiers:

Free users: Access to a basic version
Paying subscribers: More advanced outputs for ChatGPT Plus, Business, and Pro plans
Developers: OpenAI is also releasing a gpt-image-2 API, with pricing dependent on output quality and resolution

Core Model Capabilities

Text Rendering
The model accurately renders text in multiple languages, including Japanese, Korean, Hindi, Bengali, and Chinese.

Multi-Image Generation
Users can generate multiple images from a single prompt. Use cases include:

Marketing assets in various sizes
Multi-paneled comic strips
Complete documents like study booklets

Instruction Following
The model preserves requested details and renders fine-grained elements—small text, iconography, and UI elements—at up to 2K resolution.

Reasoning Capabilities
OpenAI describes the model as having "thinking capabilities," allowing it to search the web and double-check its creations. This feature requires Thinking Mode, available to paid subscribers.

Aspect Ratios & Editing

Customizable aspect ratios from 3:1 to 1:3
Editing capabilities are included, though performance in tests has shown inconsistencies compared to competing models

Generation Time: Complex images, such as multi-paneled comics, take several minutes to produce.

Performance Comparisons

Early testing shows ChatGPT Images 2.0 produces more realistic images with fewer errors than previous versions. In head-to-head comparisons with Google's Gemini model:

Feature ChatGPT Images 2.0 Google Gemini Image quality Realistic, fewer errors Comparable overall Resolution Lower than Gemini Higher resolution Text rendering Improved accuracy Standard performance Batch generation ✅ Yes — multiple images from one prompt ❌ Not available Editing Inconsistent; some distortion Preserves colors and resolution better

In one test, ChatGPT Images 2.0 generated an infographic containing accurate weather details and recognizable landmarks.

Background and Context

AI image generators using diffusion models have historically struggled with text rendering. Asmelash Teka Hadgu, founder and CEO of Lesan AI, explained in 2024 that text constitutes a small portion of image pixels, causing models to prioritize broader visual patterns.

Researchers have explored alternative mechanisms, such as autoregressive models that function more like large language models (LLMs) by predicting image content. OpenAI declined to specify the type of model powering ChatGPT Images 2.0 during a press briefing.

Important note: The model's knowledge is current through December 2025, which may affect its accuracy for prompts involving more recent events.

Industry Context

Major AI companies releasing new image models often see spikes in usage and social media trends. Last year, Google's Gemini gained popularity for hyperrealistic figurines. Earlier this year, ChatGPT Images saw viral use for AI-generated caricatures.