Skip to main content

What Image Generation Models does Pencil Support?

Written by Tim Bowers
Updated this week

Introduction

Pencil supports a range of powerful, state-of-the-art Image Generation models that can help you create visually compelling, brand-aligned content. These models are particularly effective when guided by strong prompts and visual references, including assets from your Brand Library.

Some of the use cases include:

  • Generating product imagery and lifestyle visuals for ads

  • Exploring creative directions through visual concepting

  • Iterating on design language or stylistic approaches

  • Producing on-brand visuals for social content and landing pages

  • Creating custom illustrations or graphics for campaigns

What Models Does Pencil Support?

For Image Generation, Pencil supports the following models:

  • Nano Banana 2: Gemini 3.1 Flash Image

  • Nano Banana Pro: Gemini 3

  • Nano Banana: Gemini 2.5 Flash Image

  • Google Imagen 4 Ultra

  • Google Imagen 3

  • ChatGPT Image 1.5

  • Adobe Firefly V4 Standard

  • Bria 3.2

  • Kling AI IMAGE 2.1

  • Kling AI IMAGE 2

We are always working to add to our list of supported models. Note that the models available to you may vary depending on your workspace configuration.

Brief Comparison

Each model supported by Pencil brings its own strengths. Instead of evaluating individual models in detail, it's generally more helpful to consider what kind of creative outcome you're aiming for:

  • Photorealistic or lifestyle imagery: Imagen 3, Imagen 4 Ultra, ChatGPT Image 1.5, and the Nano Banana models tend to excel at generating high-fidelity visuals for product shots, people, and everyday settings. Adobe Firefly V4 is also strong here, with added confidence around commercial licensing.

  • Product imagery and packshots: Nano Banana Pro and Nano Banana 2 are probably the strongest options for product placement and packshot-style visuals, combining high-quality output with fast generation times.

  • Multi-turn editing and grounded image creation: Nano Banana 2, Nano Banana Pro, and ChatGPT Image 1.5 support iterative editing and multi-reference inputs, making them well suited to refining a creative across several rounds.

  • Speed and rapid iteration: Nano Banana and Nano Banana 2 are optimised for fast generation cycles, producing clean outputs where responsiveness matters more than maximum fidelity.

  • Brand-safe or commercially oriented content: Bria 3.2 and Adobe Firefly V4 are designed with commercial use and brand safety in mind, offering style-consistent outputs helpful when licensing or compliance is a consideration.

  • Image expansion: Kling AI IMAGE 2.1 includes the ability to extend visuals beyond their original frame. Both Kling models are earlier-generation and available in some workspaces, but for most creative tasks the Google and OpenAI models will produce stronger results.

Summary

There’s no single best model for every use case - try a few and choose based on your creative goals. The choice of model may also matter less when aiming for specific tone and subject constraints. A richly developed Brand Library, Skilled Prompting and other conditioning assets play an incredibly important role.

Cost

Did this answer your question?