This is an advanced sandbox environment for managing workflows with neural networks. It seamlessly connects to **Groq, Google Gemini, OpenRouter, OpenAI, DeepSeek, Qwen, GLM, Claude** services, as well as local backend nodes running **Ollama/Llama.cpp**.
1. Key Setup
Enter your API key on the right and select a provider. The model list will update automatically!
2. History & Sessions
All chats are saved locally. You can switch between them using the left panel history log.
3. Code & Sandbox
When the AI generates structural code blocks, you can run and view them live inside the "Sandbox" tab component.
4. Assistant Store
Click the top right button to open the store where you can test integrated free features or access custom presets!
📎file.txt
💻
Sandbox Interpreter
🤖 Multi-Model Setup
Select multiple models to poll them simultaneously.
👥 AI Group Debate Panel
Submit your idea. Three specialized AI agents will review it from multiple perspectives (Optimist, Critic, Technologist).
🏪 Assistant Store
Get built-in standard extensions or expand library with custom presets.
FREE:
🔤 Polyglot Translator
Translates incoming texts into target language systems maintaining proper cultural idioms.
Brainstorms creative concepts, scripts, plot designs, and mechanics loops.
PAID:
🎓 Help & Provider Setup
🟢 Groq Cloud
Requires an API key from console.groq.com. Delivers the fastest inference speed on the market.
🟢 Google Gemini
Operates via official OpenAI-compliant base route. Requires an API token from Google AI Studio.
🟢 OpenRouter
Unified entry portal accessing hundreds of variants (Claude, GPT, DeepSeek). Requires token routing from openrouter.ai.
🟢 OpenAI
Standard direct infrastructure connections to mainline GPT models. Setup authorization key at platform.openai.com.
🟢 DeepSeek / Qwen / GLM
Use compatible OpenAI-style endpoints with your API key. These providers can handle multimodal prompts and rich media-aware conversations.
🟢 Claude
Anthropic's Claude models support direct API access with an Anthropic-compatible endpoint and can be used for text and media-oriented workflows.
🟢 Local Ollama
Verify your daemon is active locally (`ollama run ...`). Standard fallback routing endpoint: `http://localhost:11434/v1`.
🟢 Local Llama.cpp
Ideal when utilizing internal standard web server packaging setups. Standard fallback mapping endpoint: `http://localhost:8080/v1`.
⚠️ Please note:
When initiating queries to local servers (Ollama / Llama.cpp), your browser runtime may block transactions due to structural browser CORS security constraints. Ensure to deploy Ollama with standard origin flag variables enabled: `OLLAMA_ORIGINS="*"`.