Update README.md

README.md

@@ -1,18 +1,19 @@
 ---
 license: mit
-title: maincard
+title: rrchat AI
 sdk: static
-emoji: 📚
+emoji: ⚡
 colorFrom: red
 colorTo: indigo
 pinned: true
 short_description: Main card of rrchat!
 ---
+
 <div align="center">
-  <img src="https://huggingface.co/rrchat/Cerberus-E4B/resolve/main/cerberusbanner.png" alt="rrchat" width="100%" style="border-radius: 10px; margin-bottom: 12px;">
+  <img src="https://huggingface.co/rrchat/Cerberus-E4B/resolve/main/cerberusbanner.png" alt="rrchat AI Banner" width="100%" style="border-radius: 10px; margin-bottom: 12px;">
 
   # rrchat AI
-  **Open-source reasoning models & private mobile-ready local AI**
+  **Open-source compact reasoning models & private mobile-ready local AI**
 
   [![GitHub](https://img.shields.io/badge/GitHub-rrchat-181717?style=flat-square&logo=github)](https://github.com/leftik1/rrchat)
   [![Hugging Face](https://img.shields.io/badge/%F0%9F%A4%97%20Models-RR--Gen1-yellow?style=flat-square)](https://huggingface.co/rrchat)
@@ -21,25 +22,21 @@ short_description: Main card of rrchat!
 
 ---
 
-### 🧠 Model Series: [RR-Gen1 (Cerberus-E4B)](https://huggingface.co/collections/rrchat)
-
-First-generation multimodal reasoning models built on `gemma-4-E4B-it` (~7.5B total / 4B active LLM):
+### 🧠 Models: RR-Gen1 Series
 
-| Model | Format | Modality | Best For |
-| :--- | :---: | :---: | :--- |
-| [**Cerberus-E4B**](https://huggingface.co/rrchat/Cerberus-E4B) | LoRA Adapter | Text + Vision + Audio | Transformers, PyTorch, vLLM |
-| [**Cerberus-E4B-GGUF**](https://huggingface.co/rrchat/Cerberus-E4B-GGUF) | GGUF (`Q4_K_M`...`BF16`) | Text + Vision + Audio | llama.cpp, Ollama, LM Studio |
+Compact multimodal reasoning models with dual-mode inference (deep CoT `<think>` / instant response):
 
-* 🏆 **Math & Logic:** **60.0%** on MATH-500 (+16.7% over base Google Gemma 4).
-* ⚡ **Dual-Channel:** Native deep thought traces (`<thought>`) & direct instant responses.
-* 🛠️ **Agentic & Bilingual:** Native Gemma 4 tool-calling protocol + fluent Russian STEM reasoning.
+| Model | Base | Active Size | Modalities | Highlights | Links |
+| :--- | :--- | :---: | :---: | :--- | :---: |
+| **Mercury-1.7B** | `LFM3-VL-2B` | 1.7B | Text + Vision | Ultra-fast edge reasoning, coding (+15% HumanEval), 97% tool-use | [Model](https://huggingface.co/rrchat/Mercury-1.7B) · [GGUF](https://huggingface.co/rrchat/Mercury-1.7B-GGUF) |
+| **Cerberus-E4B** | `gemma-4-E4B-it` | 4.0B | Text + Vision + Audio | Heavy STEM reasoning (60% MATH-500), audio perception, bilingual | [Model](https://huggingface.co/rrchat/Cerberus-E4B) · [GGUF](https://huggingface.co/rrchat/Cerberus-E4B-GGUF) |
 
 ---
 
-### 📱 Client Ecosystem: [rrchat](https://github.com/leftik1/rrchat)
+### 📱 Client: [rrchat](https://github.com/leftik1/rrchat)
 
-> **Use your local PC's llama.cpp from your phone.** Clean mobile-ready AI chat interface with vision, web search, and privacy.
+Privacy-first mobile interface that connects your phone to your PC's local `llama.cpp` / `Ollama`:
 
-```bash
-# Run in Ollama:
-ollama run hf.co/rrchat/Cerberus-E4B-GGUF
+- 🔒 **100% Local & Private:** Direct local network connection, zero telemetry.
+- 👁️ **Multimodal Chat:** Vision, voice audio transcription, and document analysis.
+- 🌐 **Web Search:** Privacy-preserving agentic search without cloud lock-in.