ChatGPT Launches Images 2.0 With Enhanced Object Placement and Text Rendering.

OpenAI announced on LinkedIn that ChatGPT Images 2.0 is now available to all ChatGPT and Codex users. The update represents a significant upgrade to the platform's image generation capabilities, with improvements in instruction following, object placement, and text rendering that aim to produce more immediately usable visuals with less prompting.

The new model addresses several persistent challenges in AI image generation. Images 2.0 can now accurately place and relate multiple objects within a composition, a task that has historically required multiple attempts and careful prompt engineering. The system also renders dense text more reliably, making it more practical for creating designs that incorporate typography, labels, or written content.

OpenAI highlights that the model works across different aspect ratios without compromising quality, giving users more flexibility for various formats like social media posts, presentations, or print materials. The system also supports multilingual accuracy, generating text and visual elements correctly across different languages.

Perhaps most notably for workflow efficiency, Images 2.0 uses what OpenAI describes as "expanded visual and world knowledge" to fill in contextual gaps. This means users can provide less detailed prompts while still getting sophisticated results, as the model infers reasonable details based on broader understanding of visual concepts and real-world contexts.

ChatGPT Plus, Pro, and Business subscribers gain access to an additional feature: images with thinking. While OpenAI's announcement doesn't detail exactly how this works, the phrasing suggests the model can approach complex visual tasks with reasoning capabilities similar to those seen in recent updates to ChatGPT's agent and automation features.

This tier separation mirrors OpenAI's broader strategy of offering baseline capabilities to all users while reserving more computationally intensive features for paying subscribers.

The underlying model, called gpt-image-2, is available through OpenAI's API. This gives developers building applications on top of ChatGPT's infrastructure access to the same generation capabilities, opening possibilities for integrating the improved image generation into custom workflows, automation tools, or specialized applications.

For builders using platforms like Bubble, Glide, or Make to create NoCode applications, API access means these improvements can be incorporated into existing workflows without requiring direct interaction with ChatGPT's interface.

OpenAI says the update is available starting today for all ChatGPT users, as well as those accessing the platform through Codex. The staggered feature set based on subscription tier means free users can generate images with the new model, but advanced reasoning capabilities remain behind the paywall.

The company has not specified whether the update changes pricing, API rate limits, or generation speeds. Given the improvements in accuracy and reduced prompting requirements, users may find they can achieve desired results with fewer attempts, potentially offsetting any changes to usage costs.