1. AI Providers & API Keys
Yogatik operates on a strict Bring-Your-Own-Key (BYOK) model. Your keys are stored client-side in your browser's encrypted IndexedDB and never touch a third-party server.
- Groq: Ultra-fast inference with free tier (Llama 3.3 70B, DeepSeek R1, Mixtral).
- NVIDIA NIM: Enterprise open models with generous free cloud credits.
- Google Gemini: Gemini 2.0 Flash and Flash Thinking experimental keys.
- OpenAI & Anthropic: Direct access to GPT-4o, Claude 3.5 Sonnet, and reasoning models.
- OpenRouter: Unified gateway to 200+ frontier and specialty models.
2. 100% Offline & Local Models (Ollama)
Run models completely on your own machine with zero data leaving your device:
- Connects directly to your local Ollama daemon at
http://127.0.0.1:11434. - In the Yogatik Studio Desktop app, Ollama is discovered and spawned automatically.
- Pull, run, and benchmark models like Llama 3.2, Qwen 2.5, DeepSeek R1, and Gemma 2 offline.
3. Custom Personas & System Instructions
Tailor how the AI replies to your specific workflow:
- Prebuilt Personas: Coder, Technical Writer, Academic Researcher, Data Analyst, and Concise Assistant.
- Custom Personas: Define your own role, tone, guidelines, and tool preferences.
- Response Styles: Switch between Technical, Academic, Creative, and Explain-Like-I'm-5.
4. Data Privacy, Backup & Sync
Your conversations and workspaces belong to you:
- Local Storage: All chat transcripts, documents, and workspace files stay in local IndexedDB.
- Export Backup: Export your full history, prompts, and settings as a clean JSON archive anytime.
- Key Vault Cloud Sync (Optional): Encrypt your API keys using your Google account to sync them securely across your devices.