Model Directory
Compare image models for scientific figures
Use this page to understand which model fits your scientific figure task before you jump into Text-to-Figure, reference-based redraw, or enhancement work.
Available Models
Three model paths inside SciFig
Each model is useful, but not in the same way. Pick the one that matches the figure task in front of you.
GPT Image 2
Best when labels, annotations, prompt-following, and controlled redraw behavior matter more.
Best for text-aware figures, graphical abstracts, and edit-sensitive redraw work.
Open GPT Image 2Nano Banana Pro
Best when you want stronger default scientific-illustration quality and publication-minded visual finish.
Best for mechanism figures, publication hero visuals, and polished first-pass output.
Open Nano Banana ProNano Banana 2
Best when you want a balanced all-around model for everyday scientific figure generation tasks.
Best for repeatable day-to-day figure work across text, redraw, and enhancement modes.
Open Nano Banana 2How to Choose
What each model is best at
Think of this as a quick routing layer for scientific figure models: choose the model first, then go into the SciFig mode that matches your figure task.
GPT Image 2
- Best for text-aware figures and stronger label rendering.
- Best for controlled redraws and image-to-image editing consistency.
- Best for graphical abstracts and annotation-heavy scientific visuals.
Nano Banana Pro
- Best for stronger default scientific-illustration finish.
- Best for mechanism figures and publication hero visuals.
- Best when overall figure quality matters more than text-heavy control.
Nano Banana 2
- Best as a balanced all-around model for routine figure work.
- Best for repeatable day-to-day use across multiple modes.
- Best when you want practical speed-quality balance instead of a specialized edge.