LLM Switcher

Control AI generation with Meku’s LLM Switcher. Choose Basic, Pro, or Max profiles to balance speed, reasoning depth, and credit usage when building apps.

Choose which AI model Meku uses to generate your project. Each model has different capabilities, reasoning depth, and credit cost.

By default, Meku runs in Auto mode, selecting the most suitable model based on your prompt. You can override this at any time by picking a specific model from the switcher in the prompt bar.

Getting Started

blog image

Selecting a Model

Click the Auto button in the prompt bar to open the model dropdown. Select any model to use for your next generation. The multiplier shown next to each model indicates how many credits it uses relative to the base rate.

Switching models does not affect previously generated code. It only applies to future prompts and iterations.

Available Models

  • Auto - Meku selects the best model for your prompt automatically. Recommended for most users.
  • Gemini 3.5 Flash - 1x credits. Fast generation, good for quick prototypes and simple layouts.
  • GLM 5.2 - 1x credits. Efficient generation at base credit cost.
  • Gemini 3.1 Pro - 1.5x credits. Stronger reasoning, better for multi-page interfaces and UI consistency.
  • Claude Sonnet 5 - 2x credits. High-quality output, reliable for production-facing features.
  • GPT-5.4 - 2x credits. Strong general-purpose generation across layout and logic.
  • Claude Opus 4.8 - 3x credits. Advanced reasoning, best for complex application architecture.
  • GPT-5.5 - 3.5x credits. Highest reasoning depth for demanding generation tasks.
  • GPT-5.4 Mini - 0.5x credits. Lightweight and fast, ideal for simple edits and experiments.

Credit Multipliers

Each model applies a multiplier to your base credit usage. If a generation costs 100 base credits:

  • 0.5x - GPT-5.4 Mini consumes 50 credits
  • 1x - Gemini 3.5 Flash or GLM 5.2 consumes 100 credits
  • 1.5x - Gemini 3.1 Pro consumes 150 credits
  • 2x - Claude Sonnet 5 or GPT-5.4 consumes 200 credits
  • 3x - Claude Opus 4.8 consumes 300 credits
  • 3.5x - GPT-5.5 consumes 350 credits

When to use which Model

blog image
  • Prototyping and Quick Experiments: Use GPT-5.4 Mini, Gemini 3.5 Flash, or GLM 5.2 to explore ideas, test layouts, or run fast iterations. Low credit cost, fast output.
  • Features and UI Refinement: Use Gemini 3.1 Pro, Claude Sonnet 5, or GPT-5.4 when building stable features, refining component structure, or developing multi-page interfaces.
  • Complex Architecture: Use Claude Opus 4.8 or GPT-5.5 when your project requires deep reasoning, complex application flows, or high-precision generation.
  • Not Sure Which to Pick: Leave it on Auto. Meku picks the right model based on what you are building.

Best practices

  • Start on Auto or a 1x model to explore and prototype
  • Move to a 2x model when refining for production
  • Reserve 3x and 3.5x models for genuinely complex features where reasoning depth matters
  • Switch models freely throughout your project. It never resets or affects existing code