OpenAI's GPT Image 1.5 Brings Surgical Precision to ChatGPT's Visual Editing

What Has Actually Changed

The practical improvements span several categories that matter to professional workflows. Text rendering, long a weak point in AI image generation, has been substantially upgraded: the model now handles dense and small text with far greater accuracy, making it genuinely useful for graphic design and marketing asset production. Object manipulation, including adding, removing, or repositioning elements, now maintains compositional harmony rather than introducing visual artefacts around the edited region.

Clothing and hairstyle modifications, stylistic transformations, and conceptual blending of disparate elements have all seen meaningful gains. The model also handles complex scenes involving multiple small faces more reliably, an area where earlier versions produced outputs that required significant post-processing before they could be used professionally.

Hila Gat, Head of AI Research and Data Science at Wix, offered a direct assessment: "GPT Image 1.5 generates high-fidelity images with strong prompt adherence, preserving composition, lighting, and fine-grained detail. The results are clean, realistic, and reliable, supporting faster concept-to-production workflows."

A Dedicated Interface and What It Signals

Alongside the model upgrade, OpenAI has introduced a dedicated Images space accessible via the ChatGPT sidebar on both mobile applications and the web interface. The space offers dozens of preset filters and prompts, with the company committing to regular updates that track current design trends. The intention is clearly to reduce friction for users who have found prompt engineering a barrier to entry.

For European observers, this is worth contextualising against the regulatory environment the company is operating in. The EU AI Act classifies general-purpose AI systems with significant reach under specific transparency and copyright obligations. OpenAI's move to make image generation more accessible and embedded in everyday workflows will attract scrutiny from the European AI Office, the body established under the Act to oversee foundation model providers. Dragoș Tudorache, the Romanian MEP who was a lead architect of the AI Act, has previously argued that generative image tools require robust provenance and watermarking standards; GPT Image 1.5 will need to demonstrate compliance as European enforcement mechanisms mature through 2025 and into 2026.

On the research side, researchers at ETH Zurich's AI Centre have been examining how instruction-following improvements in diffusion-based models correlate with increases in misuse potential, particularly in synthetic media generation. The precision that makes GPT Image 1.5 useful for legitimate creative work is precisely the same capability that makes targeted manipulation of existing images more convincing. OpenAI has not yet published detailed technical documentation on the safeguards specific to this release, and European enterprise customers will want clarity on that before committing to production deployments.

Competitive Context for European Businesses

The European creative technology market is not short of alternatives. Adobe Firefly, developed partly from its European customer base and trained on licensed content, has positioned itself as the compliance-friendly option for agencies and publishers wary of copyright exposure. Stability AI, headquartered in London, continues to iterate on open-weight models. And Mistral, the Paris-based lab, is expanding its multimodal capabilities, though image generation remains a secondary priority for now.

GPT Image 1.5 enters this landscape with the considerable advantage of being embedded directly in ChatGPT, a platform with enormous penetration across European SMEs and enterprise accounts. The seamless integration means users do not need to switch tools or learn a separate interface: the upgrade is simply there, functioning across all ChatGPT models without requiring any deliberate selection.

OpenAI has also retained the previous model as a custom GPT option for users who prefer the earlier behaviour, a sensible approach that avoids disrupting established workflows whilst nudging adoption of the improved capabilities.

Key Editing Capabilities

• Clothing and style modifications: Realistic wardrobe changes and hairstyle adjustments with natural lighting preservation.
• Element manipulation: Adding, removing, or repositioning objects whilst maintaining compositional harmony.
• Stylistic transformations: Applying artistic filters and effects without losing essential image details.
• Text integration: Incorporating readable text elements with improved density and clarity handling.
• Conceptual blending: Combining disparate elements into cohesive visual narratives.
• Lighting adjustments: Modifying illumination whilst preserving natural shadow relationships.

These capabilities position ChatGPT as a credible alternative to traditional image editing software for a wide range of production tasks, not a replacement for Adobe Photoshop in complex professional contexts, but a meaningful tool for rapid iteration, concept visualisation, and content at scale.

The global rollout is live now. European users on all subscription tiers, from free through to enterprise, can access GPT Image 1.5 immediately. API availability means developers building on OpenAI's infrastructure can integrate the enhanced model into existing applications without architecture changes. The pace of iteration in this space shows no sign of slowing, and for European businesses still treating AI image generation as a novelty, that window for considered evaluation is closing fast.