A premium skincare bottle on wet glass with clean reflections, softbox highlights, pale green botanical accents, and commercial studio detail.
openai / gpt-image-2
GPT Image 2 creates polished images from prompts and reference images for product shots, characters, posters, campaign visuals, and art direction boards.
The playground uses the same request-shaped fields as the public API: prompt, image_urls, aspect_ratio, and resolution.
Pricing
GPT Image 2 costs $0.10 per generated image on sjolt. Start from text only, or add up to 9 reference images through image_urls for guided edits.
Playground Examples
Try production-style prompt starters.
A consistent character portrait for a sci-fi pilot, orange flight jacket, graphite helmet, confident expression, cinematic rim light.
A magazine cover concept for a modern architecture issue, concrete villa at dusk, warm interior light, crisp editorial typography space.
Similar Models
Compare adjacent capabilities before switching models.
Google · Image
Nano Banana Pro
GoogleGoogle's premium image generation and editing model for readable text, product mockups, visual explainers, and reference-guided brand assets.
OnlineTry model
Google · Image
Nano Banana 2
GoogleGoogle image generation and editing for photorealistic outputs, multi-reference guidance, and 1K to 4K results.
OnlineTry model
Model overview
GPT Image 2 for OpenAI image generation and reference-guided editing
GPT Image 2 brings OpenAI image generation to sjolt for high-quality text-to-image creation and reference-guided editing. Start from a prompt, add up to 9 reference images, and create consistent product shots, characters, poster concepts, social visuals, and art direction boards.
OpenAI image quality
Create refined, detailed images from natural language prompts with strong visual coherence, realistic lighting, and polished composition.
- Supports Text to image and reference-asset generation workflows.
- Fits product assets, ad previews, visual direction, and social content tests.
- Adjust configured inputs, media, and switches before generation.
Guide outputs with up to 9 reference images
Use image_urls to guide identity, pose, product form, palette, materials, or style while still directing the final result with text.
Prompt-led generation
Flexible aspect ratios and resolutions
Generate square, landscape, portrait, 3:2, and 2:3 images, with 1K, 2K, or 4K output options for ecommerce, social posts, thumbnails, editorials, and concept presentations.
- Compare available models from Google, OpenAI, ByteDance in one place.
- Copy the request body directly to the server to reduce frontend/backend parameter drift.
- Failure states, retries, result preview, and review checkpoints stay in the same workflow.
Model characteristics
Built for GPT Image 2 prompt fidelity, reference control, and production-ready output.
Prompt-faithful generation
Turn detailed instructions into coherent product shots, posters, portraits, and visual concepts with strong composition and lighting control.
Multi-reference guidance
Use up to 9 image_urls to carry over identity, product details, visual style, materials, pose, or layout from source images.
Aspect and resolution control
Generate auto, 1:1, 16:9, 9:16, 3:2, or 2:3 frames at 1K, 2K, or 4K for the target channel.
Creation and editing routes
Start from text only or provide references for guided edits while keeping the same prompt, aspect_ratio, and resolution surface.
FAQ
These are the first questions to answer when evaluating this model.
What can I make with GPT Image 2 on sjolt?
You can create high-quality images from text prompts, edit with reference images, and build practical assets such as product visuals, portraits, campaign concepts, posters, thumbnails, and social media images.
How much does GPT Image 2 cost on sjolt?
Each GPT Image 2 generation costs $0.10 per image on sjolt.
How many reference images can I upload?
You can upload up to 9 reference images with image_urls. Use them to guide identity, product details, style, material, composition, or other visual constraints.
Can GPT Image 2 work without reference images?
Yes. Use the text-to-image variant with only prompt, aspect_ratio, and resolution to generate from text.
Which aspect ratios are supported?
sjolt exposes auto, 1:1, 16:9, 9:16, 3:2, and 2:3 for GPT Image 2. auto is kept local and not forwarded upstream.