What is Race Mode
Different LLMs have unique strengths in knowledge, reasoning, and style. Race Mode allows you to run the same prompt across multiple models at the same time and compare their outputs. It increases the success rate by generating answers in parallel. You can quickly compare outputs from different LLMs to find the best result.
Now you have many choices for LLMs in Race Mode:
📃Tip: On our Choosing LLMs page, you can learn more about their differences.
How to use it
Step 1
Choose Race Mode at the bottom of the chat box on the homepage.
Step 2
Before chatting, you can try setting up other LLMs for building. The default mode is Auto. You can click Auto to customize LLMs.
Step 3
Click the gear icon. The gear icon opens a pop-up for AI Model Selector.
Step 4
Wait for a moment （5-10 minutes）, and choose the best output:
Step 5: After you choose the best option, the other options will be discarded to save storage space.
Alternative Step:
To stop Race Mode, click "Stop" at the top of the interface.
After stopping, you can:
Choose your preferred result, or
Let the Agent select one for you.
Tips:
Please note: If your project has Supabase or Stripe integration enabled, the Race Mode feature is unavailable.
Race Mode uses more credits than standard runs. Check your credit balance in advance to prevent your task from being interrupted.
Advantages of Race Mode
In complex projects, Race Mode can help you:
Work faster: Run multiple LLMs at once to test ideas quickly.
Get better results: Try different ways in parallel to improve your chances of finding the right output.
Pick the best option: Compare the strengths of different models and choose the result you like most.
Best Stages to Use Race Mode
Race Mode is especially effective in the following stages:
Exploring Possibilities
Race Mode speeds up brainstorming by generating more ideas across multiple LLMs.
Iterating prototypes
You can iterate on prototypes rapidly by fixing errors quickly with multiple LLMs.
Testing Phase
You can test which prompt, copywriting, or UI design is better. Then, you can generate multiple outputs and select the best one.
Evaluating LLMs
Run leading LLMs in parallel to see which performs best for your use case. This helps you choose the right one for your projects.
Reducing Hallucination
Race Mode lowers hallucination risks by cross-checking results with multiple LLMs.
FAQ
When should I use Race Mode?
When should I use Race Mode?
You can use Race Mode in two ways:
Same LLM, multiple versions
Run the same LLM in parallel to generate multiple results for tasks like creative ideas, code editing, or logic testing. This provides you with more diverse outputs and helps you find the best one more quickly.
Multiple LLMs running at the same time
For complex tasks, you can run different LLMs together to cover more aspects and concerns. Race Mode lets you benefit from the strengths of each model.
I am a beginner. Is Race Mode a good choice for me?
I am a beginner. Is Race Mode a good choice for me?
It depends on the complexity of your project and your goals, not your experience.
For simple projects, the usual mode is enough.
For higher accuracy, stability, or quick comparisons between LLMs, Race Mode is more effective.
Don’t worry—we’ve prepared a short beginner’s guide you can check here:
Can I use MGX as usual after I choose Race Mode?
Can I use MGX as usual after I choose Race Mode?
Yes. Simply select the best result, and then you can use MGX as usual.
⚠️ Note: If your project is integrated with Supabase or Stripe, Race Mode is not supported.
I don't have the Max Plan. Can I experience Race Mode?
I don't have the Max Plan. Can I experience Race Mode?
Sorry, Race Mode is only available to Max Plan users, since running multiple LLMs at once is costly.
Are there other limitations to Race Mode, such as the type of projects and file size?
Are there other limitations to Race Mode, such as the type of projects and file size?
No. Race Mode works for all project types and sizes.
But it consumes more Credits, so check your balance—if you run out, your project generation may fail.
Is Race Mode really much better than the usual Modes?
Is Race Mode really much better than the usual Modes?
For open-ended tasks (no single correct answer), such as copywriting, creative work, or reasoning, multiple LLMs can give you better results because of their varied outputs.
For closed tasks (such as fact-checking or data retrieval) that require a single correct answer, different LLMs often yield the same result, so the benefit is smaller.