OpenAI releases ChatGPT Images 2.0 with integrated reasoning and text-image composition

OpenAI has released ChatGPT Images 2.0, adding reasoning capabilities to image generation for the first time. The model can now generate context-aware infographics and complex visual layouts by combining data retrieval, reasoning, and visual composition.

Thinking mode enables complex workflows

The most significant change is the integration of OpenAI's thinking capabilities into image generation. When given a vague prompt like "Generate an infographic about activities I should do with tomorrow's weather in San Francisco in mind," Images 2.0 will gather weather data, determine appropriate activities, and produce coherent visual outputs.

According to OpenAI, this shifts image generation from creating "decorations" to functioning as "a visual language" where "a good image does what a good sentence does -- it selects, arranges, and reveals."

The model can generate multiple images per prompt with continuity across outputs, treating visual design as a reasoning task rather than pure generation.

Technical specifications

Images 2.0 supports aspect ratios from 3:1 (wide) to 1:3 (tall), addressing a long-standing limitation where ChatGPT would often ignore aspect ratio requests. The model outputs at up to 2K resolution with what OpenAI claims is improved accuracy for object placement, text rendering, and complex compositions.

The model handles small text, UI elements, and stylistic constraints, though early testing revealed inconsistent results with brand fidelity. ZDNET's pre-release testing found the model struggled to accurately reproduce logos, generating outdated versions or adding spurious design elements.

Availability and pricing

Images 2.0 is available starting April 21, 2026 to all ChatGPT and Codex users. The advanced reasoning capabilities require ChatGPT Plus, Pro, Business, or Enterprise subscriptions. Users must select "Thinking" from the ChatGPT dropdown menu to access reasoning features.

At launch, the desktop version only is available, with mobile support coming later. The model is accessible via API using the gpt-image-2 endpoint. According to OpenAI, API pricing varies based on quality settings, thinking depth, and output resolution, though specific per-image pricing was not disclosed.

What this means

This release represents OpenAI's attempt to position image generation as a reasoning task rather than pure pixel synthesis. By integrating its thinking capabilities, the model can theoretically handle end-to-end visual design workflows—from gathering data to composing final layouts. However, early testing suggests precision issues with brand elements remain, which could limit enterprise adoption where visual consistency is critical. The API access via gpt-image-2 indicates OpenAI is positioning this for integration into design tools and automated content pipelines.