Skip to content
TheChatArchiveThe Chat Archive
Qwenqwen3-30b-a3b· Alibaba (Qwen)Hugging Face LMArena Human Preference 140K

find (reddit forums) what custom models i can add to my cursor, and they will work?…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-opus-4-20250514’s answer.
  1. User

    find (reddit forums) what custom models i can add to my cursor, and they will work? additionally i needto know what model is the best across gemini-2.5-flash-preview, gemini 2.5 pro and gemini 2.5 pro-exp-03.25

  2. Qwen Assistant

    To address your query, here's a structured breakdown of the information you need:


    1. Custom Models for Cursor (AI Code Editor)

    Cursor is an AI-powered code editor that primarily integrates with Google's Gemini models (e.g., Gemini 2.5 Flash, Gemini 2.5 Pro, etc.) and other LLMs via APIs. However, Cursor does not natively support adding arbitrary custom models (like self-hosted models or third-party models not explicitly supported by Cursor). Here's what you can do:

    Options for Custom Models:

    • Use APIs for External Models:
      If you have a custom model (e.g., hosted on Hugging Face, Ollama, or your own server), you can use Cursor's "Custom API" feature to connect to it. This requires:

      • A public API endpoint for your model.
      • Configuration of the API key and parameters in Cursor's settings.
      • Example: You could integrate a local model like Llama 3 via Ollama or a Hugging Face API.
    • Check for Community Plugins:
      Some users create plugins or integrations for Cursor to support custom models. Check platforms like GitHub or the Cursor Discord community for such tools.

    • Limitations:
      Cursor is optimized for Google's models (Gemini) and may not support all third-party models out of the box. Performance, latency, or compatibility issues could arise with non-Google models.


    2. Best Gemini Model: Gemini 2.5 Flash Preview vs. Gemini 2.5 Pro vs. Gemini 2.5 Pro-Exp-03.25

    Here's a comparison of the three models based on performance, use cases, and stability:

    A. Gemini 2.5 Flash Preview

    • Strengths:
      • Lightweight and fast (ideal for quick tasks like code suggestions, simple queries).
      • Lower cost for API usage.
    • Weaknesses:
      • Less accurate for complex reasoning or long-context tasks.
      • Limited to basic coding and general knowledge.
    • Best For:
      • Rapid prototyping, small code snippets, or low-latency interactions.

    B. Gemini 2.5 Pro

    • Strengths:
      • High accuracy for complex tasks (e.g., multi-step reasoning, code generation, debugging).
      • Better at handling long contexts and technical queries.
      • More robust for enterprise-grade applications.
    • Weaknesses:
      • Higher cost and slower response times compared to Flash.
    • Best For:
      • Advanced coding, detailed problem-solving, or critical workflows where accuracy is essential.

    C. Gemini 2.5 Pro-Exp-03.25 (Experimental)

    • Strengths:
      • Likely includes new features (e.g., improved reasoning, multilingual support, or specialized training).
      • May offer better performance for niche tasks (e.g., scientific computing, domain-specific knowledge).
    • Weaknesses:
      • Unstable (not recommended for production use).
      • Potential bugs or incomplete features.
    • Best For:
      • Testing cutting-edge capabilities or specific research/development scenarios.

    3. Recommendation

    • For Most Users:
      Use Gemini 2.5 Pro for its balance of accuracy and performance. It’s ideal for code generation, debugging, and complex tasks.
    • For Speed/Cost:
      Use Gemini 2.5 Flash Preview for lightweight tasks.
    • For Experimental Use:
      Try Gemini 2.5 Pro-Exp-03.25 if you need access to new features, but be prepared for potential instability.

    4. Steps to Test Models in Cursor

    1. Check Cursor’s Model Settings:
      Go to Settings > AI Assistant > Model to see available options.
    2. Switch Between Models:
      If your plan allows, toggle between Flash, Pro, and experimental models.
    3. Test with Your Workflow:
      Run code generation, debugging, or problem-solving tasks to evaluate performance.

    5. Additional Notes

    • Model Availability:
      Some models (like Pro-Exp-03.25) may require a Google Cloud API key and access to experimental features.
    • Cost:
      Pro models are more expensive than Flash. Check Cursor’s pricing or Google’s API pricing for details.

    If you’re unsure about the best model for your use case, start with Gemini 2.5 Pro and test with your specific tasks. For custom models, ensure your API setup is compatible with Cursor’s requirements.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.