Update README.md

README.md

@@ -130,18 +130,15 @@ The `notes` array in each aspect section contains warnings such as:
 
 ## Setup
 
-Space Settings → **Repository Secrets** → add `HF_TOKEN`  
-(free Read token from [huggingface.co/settings/tokens](https://huggingface.co/settings/tokens))
+Space Settings → **Repository Secrets** → add `GEMINI_API_KEY`
+
+Get a **free** Gemini API key at [aistudio.google.com/apikey](https://aistudio.google.com/apikey)  
+No credit card required — free tier includes **1,500 requests/day**.
 
 ---
 
-## Models (free tier, tried in order)
+## Models (Google Gemini, free tier)
 
-1. `google/gemma-4-27b-it` — primary (best quality)
-2. `meta-llama/Llama-4-Scout-17B-16E-Instruct` — Llama 4 Scout vision
-3. `Qwen/Qwen2.5-VL-7B-Instruct` — dedicated vision model
-4. `Qwen/Qwen2-VL-7B-Instruct` — previous Qwen VL, very stable
-5. `meta-llama/Llama-3.2-11B-Vision-Instruct` — reliable fallback
-6. `microsoft/Phi-3.5-vision-instruct` — lightweight fallback
-7. `HuggingFaceM4/idefics3-8b-llama3` — HF native, always available
-8. `mistralai/Pixtral-12B-2409` — Mistral vision model
+1. `gemini-2.0-flash` — primary (best quality + speed, free)
+2. `gemini-1.5-flash` — fallback 1 (previous gen, very stable)
+3. `gemini-1.5-flash-8b` — fallback 2 (lightest, always available)