Changing the base AI model that Leetcode Wizard uses

Leetcode Wizard offers over 220 different AI models that can be used as the base model to generate the final output. You can change the model used by each individual prompt from the "Edit prompts" page (Profile dropdown (top left) -> "Edit prompts").

Recommended models

GPT-5-2 by OpenAI

This is the default and most tested model that we currently offer. It will respond very quickly (within a few seconds) and has a very stable, predictable output. We recommend this for most users.

Claude 4.5 Sonnet by Anthropic

The other model we recommend is Claude 4.5 Sonnet. Claude 4.5 is great at coding and offers results that are on par with OpenAI's models.

DeepSeek-V3.2

The newest DeepSeek model is a game changer for Leetcode interviews. In our tests, this model has shown significant improvements over other models.

Benchmarks

To know which models will perform best at any time, we recommend checking out the SWE-bench Leaderboard. This leaderboard keeps track of how well certain AI models are doing at coding problems.