Increase max tokens to 1024

Fix model name to gemini-2.5-flash

Migrate to google-genai SDK with gemini-2.5-flash

Switch to gemini-2.5-flash-preview-05-20

Switch to Gemini 2.0 Flash API via middleman — no model loading

Switch to Qwen2.5-0.5B-Instruct for faster CPU inference

Fix apply_chat_template BatchEncoding compatibility

Add portfolio chat agent (FastAPI + Qwen2.5-1.5B-Instruct)

initial commit