Choosing a Text Model
The text model is the AI that writes your video's script. When you type a prompt describing your video idea, the text model turns that description into a structured, narrated script with scenes, hooks, and calls to action.
Written By Rishikesh from ngram
Last updated About 1 month ago
Choosing the right text model affects the quality, creativity, and speed of your script.
Where to Find It
The text model selector is located in the settings bar below the prompt box on the home page. Look for the button labeled with the current model name (e.g., "Gemini 3 Pro").

Click the button to open a dropdown menu showing all available providers and models.
Available Providers and Models
Google (Default)
Google offers several models through the Gemini family. Google is the default provider because it delivers a strong balance of quality, speed, and cost.

When you hover over or click "Google" in the dropdown, a submenu appears with all the Gemini model variants.
Anthropic
Anthropic provides Claude models known for clear, well-structured writing. If you want scripts that are precise and follow instructions closely, Anthropic is a strong choice.
OpenAI
OpenAI offers GPT models that are widely used and versatile. These models are effective for a broad range of video styles and topics.
Groq
Groq specializes in extremely fast inference. If speed is your top priority and you want a script generated as quickly as possible, Groq is worth trying. The tradeoff is that the output may be less polished compared to Pro-tier models from other providers.
Understanding Pro vs. Flash
You will notice that some Google models include "Pro" or "Flash" in their names. Here is what that means:
Pro models are the most capable. They produce higher-quality scripts with better structure, more creative language, and stronger adherence to your prompt. They take a bit longer and use more credits.
Flash models are optimized for speed. They generate scripts faster and use fewer credits, but the output may be slightly less refined. Flash models are great for quick iterations or when you plan to edit the script yourself.
How the Text Model Affects Your Video
The text model only controls the script-writing step. It does not affect:
The images generated for your storyboard (that is the image model)
The video animation (that is the video model)
The voiceover (that is the voice setting)
However, a better script leads to a better video overall. The script determines:
Scene structure -- how many scenes and how they flow
Narration text -- the words the voiceover will speak
Scene descriptions -- the visual descriptions that guide image generation
Pacing -- how long each scene lasts relative to others
When to Change the Text Model
Stick with the default (Gemini 3 Pro) if you are just getting started. It works well for most use cases.
Consider switching if:
You need faster results: Switch to Gemini 3 Flash or Groq
You want the highest quality script: Try Gemini 2.5 Pro
You prefer a specific writing style: Experiment with Anthropic or OpenAI
You are iterating quickly on ideas: Use a Flash model for drafts, then switch to Pro for the final version
Tips for Getting the Best Script
Be specific in your prompt. The text model performs better when you describe your audience, tone, and key points clearly.
Enable Deep Research (in the Tools menu) if you want the AI to look up information about your product or topic before writing. This produces more accurate scripts.
Use the default model first, then experiment. You can always regenerate the script with a different model without losing your other settings.
Review and edit. Even the best AI model benefits from a human review. Use the Edit button on the script panel to make adjustments.
Credit Usage
Different models consume different amounts of credits:
Pro models use more credits per script generation
Flash models use fewer credits per script generation
Groq models are generally the most credit-efficient
The exact credit cost depends on the length and complexity of your script. You can see credits used during your session at the top of the chat page.
Next Steps
Choosing an Image Model -- select the AI that generates your scene images
Setting Video Duration and Aspect Ratio -- control how long your video is and its dimensions
Tools: Animation Mode and Deep Research -- enable Deep Research for better scripts