import gradio as gr with gr.Blocks(theme=gr.themes.Soft(primary_hue="blue"), title="Quick Start") as demo: gr.Markdown(""" # 🚀 dispatchAI Quick Start Guide ## Step 1: Install ```bash pip install dispatchai[gguf] ``` ## Step 2: Load a Model ```python from dispatchai import load_model model = load_model("SmolLM2-135M-Instruct-mobile", backend="gguf") ``` ## Step 3: Chat! ```python print(model.chat("What is the capital of France?")) # → "The capital of France is Paris." ``` ## Step 4: Find the Right Model ```python from dispatchai import recommend rec = recommend(ram_mb=2048, task="chat") print(f"Best for 2GB RAM: {rec['recommended']['name']}") ``` ## Step 5: Check Savings ```python from dispatchai import calculate_cost result = calculate_cost(daily_queries=10000) print(f"Save ${result['savings']}/year vs cloud") ``` ## Verified Models - ✅ 31 models fully working - ❌ 0 broken - ⚠️ 0 partial - 📱 3 phone-verified on Snapdragon 865 ## Top 3 Models | Model | Size | Speed | Phone Speed | |-------|------|-------|-------------| | SmolLM2-135M | 101MB | 59.7 t/s | 46.0 t/s | | Qwen2.5-0.5B-int4 | 469MB | 12.5 t/s | 23.2 t/s | | Llama-3.2-1B-Q4 | 770MB | 11.3 t/s | 5.4 t/s | ## Links - [All Models](https://huggingface.co/dispatchAI) - [SDK](https://huggingface.co/dispatchAI/dispatchAI-SDK) - [MCP Hub](https://huggingface.co/spaces/dispatchAI/mcp-hub) - [Leaderboard](https://huggingface.co/spaces/dispatchAI/mobile-llm-leaderboard) - [Hub](https://huggingface.co/spaces/dispatchAI/hub) --- Dispatch AI (FZE) — Sharjah, UAE. License No. 10818. 🚀 [dispatchAI](https://huggingface.co/dispatchAI) — Small. Mobile. Free. UAE-built. """) demo.launch()