Introduction
Pencil supports several of the best, state-of-the-art Text Generation models that can help you generate creative brand-safe copy. These models are powerful tools, especially when drawing from your Brand Library, and used with skilled Prompting. Some of the use cases include:
Generating and iterating taglines, CTAs and other short-form copy
Generating strategic-copy - audience personas, customer insights and targeted suggestions
Generating creative product descriptions in both long-form and short-form contexts.
Translation of copy into a range of languages
Generating copy for email campaigns and blog posts
What Models Does Pencil Support?
For Text Generation Pencil supports the following models:
OpenAI GPT-4o
Gemini 2.0 Flash
We are always working to add to our list of supported models.
Brief Comparison
OpenAI GPT 4o:
Generally strong across a wide range of use cases, including nuanced tone and brand-safe outputs
Often favoured for more structured or context-sensitive copy
Can sometimes produce longer, more verbose results by default
Particularly effective when paired with a well-structured prompt and clear input context
Supports text, image and audio inputs
Offers a human-like conversational experience - natural and engaging
Gemini 2.0 Flash
Very fast generation speeds, making it well suited for rapid iteration and testing.
Tends to produce punchier and more concise outputs, which can be useful for short-form copy
Can require more guidance to maintain tone consistency across outputs
An effective option for exploration of divergent copy ideas at speed.
Gemini 2.5 Flash
In comparison to Gemini 2.0 Flash, Gemini 2.5 Flash tends to perform better with deeper reasoning or context-heavy work.
It is better equipped for more complex tasks with large inputs.
Summary
There’s no single 'best' model—each has its strengths. Choose based on what you’re trying to achieve, and feel free to switch as your needs evolve. You can switch between models at any point, depending on whether you’re exploring widely or refining towards a final version.