Dropstone AI Models
This section provides a comprehensive overview of Dropstone AI models and instructions for using Open Source Models (Unlimited).
Dropstone AI Models provide developers and teams with advanced tools to accelerate coding, debugging, and code analysis. This documentation explains the available models, subscription plan availability, and step-by-step instructions for using Open Source Models (Unlimited).
Dropstone Models (Fast)
Fast models are optimized for real-time performance and high-quality AI-assisted coding.
| Model | Credits | Free | Pro | Teams | Enterprise | Trial |
|---|---|---|---|---|---|---|
| gpt-4o-mini | 0 | ✔ | ✔ | ✔ | ✔ | ✔ |
| claude-3-haiku | 0 | ✔ | ✔ | ✔ | ✔ | ✔ |
| gpt-3.5-turbo | 0 | ✔ | ✔ | ✔ | ✔ | ✔ |
| deepseek-v3 | 0 | ✔ | ✔ | ✔ | ✔ | ✔ |
| gemini-2.5-flash | 2 | ✔ | ✔ | ✔ | ✔ | |
| claude-3-sonnet | 7 | ✔ | ✔ | ✔ | ✔ | |
| claude-3.5-sonnet | 8 | ✔ | ✔ | ✔ | ✔ | |
| gemini-2.5-pro | 9 | ✔ | ✔ | ✔ | ✔ | |
| gpt-4o | 10 | ✔ | ✔ | ✔ | ✔ | |
| o3-mini | 10 | ✔ | ✔ | ✔ | ✔ | |
| claude-4-sonnet | 12 | ✔ | ✔ | ✔ | ✔ | |
| gpt-5 | 14 | ✔ | ✔ | ✔ | ✔ | |
| claude-4-sonnet:thinking | 15 | ✔ | ✔ | ✔ | ✔ | |
| o3:thinking | 15 | ✔ | ✔ | ✔ | ✔ | |
| claude-4.5-sonnet | 18 | ✔ | ✔ | ✔ | ✔ | |
| claude-opus-4 | 150 | ✔ | ✔ | ✔ | ✔ | |
| claude-opus-4.1 | 150 | ✔ | ✔ | ✔ | ✔ |
Notes:
- Credits indicate usage cost per request.
- Pro users get 900 credits per month.
- Free plan users have unlimited access to 0-credit models.
- Teams and Enterprise users have unlimited credits.
Open Source Models (Unlimited)
Open-source models can be installed locally on your PC and used without any credit cost. Performance depends on your hardware specifications (CPU/GPU, RAM).
Any open-source model can be used by installing it on your PC through Dropstone. Popular options include:
| Model Family | Example Models | Requirements | Free | Pro | Teams | Enterprise |
|---|---|---|---|---|---|---|
| Qwen-Coder Series | Qwen2.5-Coder 7B/14B/32B | 8GB+ RAM | ✔ | ✔ | ✔ | ✔ |
| CodeLlama | CodeLlama 7B/13B/34B | 8GB+ RAM | ✔ | ✔ | ✔ | ✔ |
| DeepSeek-Coder | DeepSeek-Coder 1.3B/6.7B/33B | 4GB+ RAM | ✔ | ✔ | ✔ | ✔ |
| StarCoder | StarCoder/StarCoder2 | 16GB+ RAM | ✔ | ✔ | ✔ | ✔ |
| Mistral | Mistral 7B/Mixtral 8x7B | 8GB+ RAM | ✔ | ✔ | ✔ | ✔ |
Notes:
- All open-source models cost 0 credits once installed locally.
- Model performance depends on your PC hardware (faster GPU = better performance).
- Available to all users (Free, Pro, Teams, Enterprise).
- You can install any compatible open-source model from HuggingFace or other sources.
Agent Request Limits
Dropstone includes powerful AI agent capabilities with the following usage limits:
| Plan | Agent Requests per Month | Credits per Month |
|---|---|---|
| Free | 50 requests | N/A (0-credit models only) |
| Pro | Unlimited | 900 credits |
| Teams | Unlimited | Unlimited credits |
| Enterprise | Unlimited | Unlimited credits |
Notes:
- Free users: Limited to 50 agent requests per month, but can use unlimited 0-credit models.
- Pro users: Unlimited agent requests, limited by 900 monthly credits.
- Teams & Enterprise: Unlimited agent requests and unlimited credits.
- Agent requests reset monthly on your subscription date.
Using Open Source Models (Unlimited)
Open-source models can be installed locally and used without restrictions. The following steps outline installation, configuration, and usage.
Step 1: Check Model Availability
- Open the Dropstone Dashboard and navigate to the Models section.
- Identify models marked Not Installed.
- Select the model to install, such as Qwen3 32B.
Ensure your system meets the GPU/CPU requirements for the selected model.
Step 2: Install the Model
- Click Install next to the selected model.
- Dropstone will download the model weights and dependencies.
- Monitor installation progress; large models may take significant time.
Ensure sufficient disk space and memory.
Step 3: Configure the Model
- Click Configure after installation.
- Set the following parameters:
- Device: GPU (recommended) or CPU
- Batch Size: Adjust based on available memory
- Max Tokens: Limit the output length
- Save the configuration.
Advanced users can fine-tune the model using project-specific datasets.
Step 4: Load the Model
- Click Load Model in the dashboard.
- Wait for initialization; large models may take several minutes.
- Confirm the model status shows Ready.
Step 5: Using the Model
- Open Dropstone Workspace or Code Analysis Tool.
- Select the installed open-source model.
- Enter your prompt or code snippet.
- Click Run/Analyze to generate output.
- Review results and apply suggestions to your code.
Use smaller models for fast testing, and larger models for comprehensive codebase analysis.
Step 6: Manage Model Usage
- Monitor resource consumption via the dashboard.
- Regularly update models to access improvements.
- Uninstall unused models to free disk space.
Step 7: Troubleshooting
| Issue | Solution |
|---|---|
| Model fails to load | Verify RAM/GPU availability and reduce batch size. |
| Slow performance | Use a smaller model or reduce max tokens. |
| Installation fails | Ensure stable internet and sufficient disk space. |
This documentation provides a professional, structured reference for Dropstone AI Models and the Open Source Models (Unlimited) workflow, suitable for developers, teams, and enterprise environments.
`