We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
OpenAI Blog·Dec 16·3 min read
The new ChatGPT Images is here
TL;DR
Image generation models previously faced limitations in edit precision, detail consistency, and generation speed. OpenAI released GPT-Image-1.5, an upgraded flagship image generation model available in ChatGPT and via API.
✦ Why It Matters
Engineers can now integrate faster, more controllable image generation into applications with improved consistency and lower latency costs.
Key Takeaways
- Integrate GPT-Image-1.5 via ChatGPT or API for production image workflows requiring speed and precision.
- Leverage 4× faster generation to reduce latency in real-time creative or design applications.
- Use improved edit precision for targeted image modifications without full regeneration.
- Evaluate cost-per-inference improvements when migrating from earlier generation models.
Full Summary
Image generation has become a core capability in AI assistants, but earlier models struggled with precise editing control, maintaining consistent details across variations, and generation latency. OpenAI developed GPT-Image-1.5, an improved image generation model designed to address these constraints.
The model achieves up to 4× faster inference speed compared to its predecessor, enabling near-real-time image creation workflows. It also enhances edit precision—allowing users to make targeted modifications to specific image regions—and improves detail consistency, ensuring visual elements remain coherent across multiple generations.
The upgrade rolls out to all ChatGPT users and is simultaneously available through OpenAI's API, enabling both consumer and developer access. These improvements reduce friction in creative workflows and lower computational costs for production applications.