Choosing a model depends a lot on the specific task and command. Rewritebar, therefore, allows you to select a specific provider and model at the command level. My rule of thumb is that I use GPT-4.1-mini as the default model for most commands and then select a smarter one for commands that require it.
| Use Case |
Example Commands |
Models |
| Rewriting a text sentence where the model does not need a deep understanding of the context |
- Fix Grammar & Spelling |
|
- Translate in another language
- Change Tone
- Transform into title or lowercase | Small-size models are, in most cases, a good fit here, as they are smart enough but also fast and cheap.
Models:
- GPT-4o-mini
- GPT-4.1-mini
- lama-3.2-8B
- Gemini-2.5-flash |
| Summarization of long text | - Summarize
- Make Text Shorter | Use midsize models like GPT-4o or GPT-4.1. |
| Improving Prompts | - Improve Prompt
- Generate Prompt | Use mid-size models like GPT-4o or GPT-4.1. |
| The model should extend or invent more content. | - Make Text Longer | Use smarter models like GPT-4o or higher, depending on the topic. |
Further Consideration
- I mainly use OpenAI models during the development of RewriteBar. So, prompts are mainly optimized for these models (GPT-4x family). Consider refining the prompt if you run into issues with your selected model (e.g., local models).