Update app.py

app.py

@@ -6,6 +6,8 @@ import base64
 from PIL import Image
 import io
 import logging
+import PyPDF2
+import markdown
 
 # Configure logging
 logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
@@ -14,82 +16,80 @@ logger = logging.getLogger(__name__)
 # API key
 OPENROUTER_API_KEY = os.environ.get("OPENROUTER_API_KEY", "")
 
-# Model list with context sizes
+# Model list with context sizes - organized by category
 MODELS = [
     # Vision Models
-    ("Meta: Llama 3.2 11B Vision Instruct (free)", "meta-llama/llama-3.2-11b-vision-instruct:free", 131072),
-    ("Qwen: Qwen2.5 VL 72B Instruct (free)", "qwen/qwen2.5-vl-72b-instruct:free", 131072),
-    ("Qwen: Qwen2.5 VL 32B Instruct (free)", "qwen/qwen2.5-vl-32b-instruct:free", 8192),
-    ("Qwen: Qwen2.5 VL 7B Instruct (free)", "qwen/qwen-2.5-vl-7b-instruct:free", 64000),
-    ("Qwen: Qwen2.5 VL 3B Instruct (free)", "qwen/qwen2.5-vl-3b-instruct:free", 64000),
+    {"category": "Vision", "models": [
+        ("Meta: Llama 3.2 11B Vision Instruct", "meta-llama/llama-3.2-11b-vision-instruct:free", 131072),
+        ("Qwen2.5 VL 72B Instruct", "qwen/qwen2.5-vl-72b-instruct:free", 131072),
+        ("Qwen2.5 VL 32B Instruct", "qwen/qwen2.5-vl-32b-instruct:free", 8192),
+        ("Qwen2.5 VL 7B Instruct", "qwen/qwen-2.5-vl-7b-instruct:free", 64000),
+        ("Qwen2.5 VL 3B Instruct", "qwen/qwen2.5-vl-3b-instruct:free", 64000),
+    ]},
     
     # Gemini Models
-    ("Google: Gemini Pro 2.0 Experimental (free)", "google/gemini-2.0-pro-exp-02-05:free", 2000000),
-    ("Google: Gemini Pro 2.5 Experimental (free)", "google/gemini-2.5-pro-exp-03-25:free", 1000000),
-    ("Google: Gemini 2.0 Flash Thinking Experimental 01-21 (free)", "google/gemini-2.0-flash-thinking-exp:free", 1048576),
-    ("Google: Gemini Flash 2.0 Experimental (free)", "google/gemini-2.0-flash-exp:free", 1048576),
-    ("Google: Gemini Flash 1.5 8B Experimental", "google/gemini-flash-1.5-8b-exp", 1000000),
-    ("Google: Gemini 2.0 Flash Thinking Experimental (free)", "google/gemini-2.0-flash-thinking-exp-1219:free", 40000),
-    ("Google: LearnLM 1.5 Pro Experimental (free)", "google/learnlm-1.5-pro-experimental:free", 40960),
+    {"category": "Gemini", "models": [
+        ("Gemini Pro 2.0 Experimental", "google/gemini-2.0-pro-exp-02-05:free", 2000000),
+        ("Gemini Pro 2.5 Experimental", "google/gemini-2.5-pro-exp-03-25:free", 1000000),
+        ("Gemini 2.0 Flash Thinking Experimental", "google/gemini-2.0-flash-thinking-exp:free", 1048576),
+        ("Gemini Flash 2.0 Experimental", "google/gemini-2.0-flash-exp:free", 1048576),
+        ("Gemini Flash 1.5 8B Experimental", "google/gemini-flash-1.5-8b-exp", 1000000),
+        ("LearnLM 1.5 Pro Experimental", "google/learnlm-1.5-pro-experimental:free", 40960),
+    ]},
     
     # Llama Models
-    ("Meta: Llama 3.3 70B Instruct (free)", "meta-llama/llama-3.3-70b-instruct:free", 8000),
-    ("Meta: Llama 3.2 3B Instruct (free)", "meta-llama/llama-3.2-3b-instruct:free", 20000),
-    ("Meta: Llama 3.2 1B Instruct (free)", "meta-llama/llama-3.2-1b-instruct:free", 131072),
-    ("Meta: Llama 3.1 8B Instruct (free)", "meta-llama/llama-3.1-8b-instruct:free", 131072),
-    ("Meta: Llama 3 8B Instruct (free)", "meta-llama/llama-3-8b-instruct:free", 8192),
-    ("NVIDIA: Llama 3.1 Nemotron 70B Instruct (free)", "nvidia/llama-3.1-nemotron-70b-instruct:free", 131072),
+    {"category": "Llama", "models": [
+        ("Llama 3.3 70B Instruct", "meta-llama/llama-3.3-70b-instruct:free", 8000),
+        ("Llama 3.2 3B Instruct", "meta-llama/llama-3.2-3b-instruct:free", 20000),
+        ("Llama 3.2 1B Instruct", "meta-llama/llama-3.2-1b-instruct:free", 131072),
+        ("Llama 3.1 8B Instruct", "meta-llama/llama-3.1-8b-instruct:free", 131072),
+        ("Llama 3 8B Instruct", "meta-llama/llama-3-8b-instruct:free", 8192),
+        ("Llama 3.1 Nemotron 70B Instruct", "nvidia/llama-3.1-nemotron-70b-instruct:free", 131072),
+    ]},
     
     # DeepSeek Models
-    ("DeepSeek: DeepSeek R1 Zero (free)", "deepseek/deepseek-r1-zero:free", 163840),
-    ("DeepSeek: R1 (free)", "deepseek/deepseek-r1:free", 163840),
-    ("DeepSeek: DeepSeek V3 Base (free)", "deepseek/deepseek-v3-base:free", 131072),
-    ("DeepSeek: DeepSeek V3 0324 (free)", "deepseek/deepseek-v3-0324:free", 131072),
-    ("DeepSeek: DeepSeek V3 (free)", "deepseek/deepseek-chat:free", 131072),
-    ("DeepSeek: R1 Distill Qwen 14B (free)", "deepseek/deepseek-r1-distill-qwen-14b:free", 64000),
-    ("DeepSeek: R1 Distill Qwen 32B (free)", "deepseek/deepseek-r1-distill-qwen-32b:free", 16000),
-    ("DeepSeek: R1 Distill Llama 70B (free)", "deepseek/deepseek-r1-distill-llama-70b:free", 8192),
+    {"category": "DeepSeek", "models": [
+        ("DeepSeek R1 Zero", "deepseek/deepseek-r1-zero:free", 163840),
+        ("DeepSeek R1", "deepseek/deepseek-r1:free", 163840),
+        ("DeepSeek V3 Base", "deepseek/deepseek-v3-base:free", 131072),
+        ("DeepSeek V3 0324", "deepseek/deepseek-v3-0324:free", 131072),
+        ("DeepSeek V3", "deepseek/deepseek-chat:free", 131072),
+        ("DeepSeek R1 Distill Qwen 14B", "deepseek/deepseek-r1-distill-qwen-14b:free", 64000),
+        ("DeepSeek R1 Distill Qwen 32B", "deepseek/deepseek-r1-distill-qwen-32b:free", 16000),
+        ("DeepSeek R1 Distill Llama 70B", "deepseek/deepseek-r1-distill-llama-70b:free", 8192),
+    ]},
     
-    # Gemma Models
-    ("Google: Gemma 3 27B (free)", "google/gemma-3-27b-it:free", 96000),
-    ("Google: Gemma 3 12B (free)", "google/gemma-3-12b-it:free", 131072),
-    ("Google: Gemma 3 4B (free)", "google/gemma-3-4b-it:free", 131072),
-    ("Google: Gemma 3 1B (free)", "google/gemma-3-1b-it:free", 32768),
-    ("Google: Gemma 2 9B (free)", "google/gemma-2-9b-it:free", 8192),
+    # Other Popular Models
+    {"category": "Other Popular Models", "models": [
+        ("Mistral Nemo", "mistralai/mistral-nemo:free", 128000),
+        ("Mistral Small 3.1 24B", "mistralai/mistral-small-3.1-24b-instruct:free", 96000),
+        ("Gemma 3 27B", "google/gemma-3-27b-it:free", 96000),
+        ("Gemma 3 12B", "google/gemma-3-12b-it:free", 131072),
+        ("Gemma 3 4B", "google/gemma-3-4b-it:free", 131072),
+        ("DeepHermes 3 Llama 3 8B Preview", "nousresearch/deephermes-3-llama-3-8b-preview:free", 131072),
+        ("Qwen2.5 72B Instruct", "qwen/qwen-2.5-72b-instruct:free", 32768),
+    ]},
     
-    # Mistral Models
-    ("Mistral: Mistral Nemo (free)", "mistralai/mistral-nemo:free", 128000),
-    ("Mistral: Mistral Small 3.1 24B (free)", "mistralai/mistral-small-3.1-24b-instruct:free", 96000),
-    ("Mistral: Mistral Small 3 (free)", "mistralai/mistral-small-24b-instruct-2501:free", 32768),
-    ("Mistral: Mistral 7B Instruct (free)", "mistralai/mistral-7b-instruct:free", 8192),
-    
-    # Qwen Models
-    ("Qwen: Qwen2.5 72B Instruct (free)", "qwen/qwen-2.5-72b-instruct:free", 32768),
-    ("Qwen: QwQ 32B (free)", "qwen/qwq-32b:free", 40000),
-    ("Qwen: QwQ 32B Preview (free)", "qwen/qwq-32b-preview:free", 16384),
-    ("Qwen2.5 Coder 32B Instruct (free)", "qwen/qwen-2.5-coder-32b-instruct:free", 32768),
-    ("Qwen 2 7B Instruct (free)", "qwen/qwen-2-7b-instruct:free", 8192),
-    
-    # Other Models
-    ("Nous: DeepHermes 3 Llama 3 8B Preview (free)", "nousresearch/deephermes-3-llama-3-8b-preview:free", 131072),
-    ("Moonshot AI: Moonlight 16B A3B Instruct (free)", "moonshotai/moonlight-16b-a3b-instruct:free", 8192),
-    ("Microsoft: Phi-3 Mini 128K Instruct (free)", "microsoft/phi-3-mini-128k-instruct:free", 8192),
-    ("Microsoft: Phi-3 Medium 128K Instruct (free)", "microsoft/phi-3-medium-128k-instruct:free", 8192),
-    ("OpenChat 3.5 7B (free)", "openchat/openchat-7b:free", 8192),
-    ("Reka: Flash 3 (free)", "rekaai/reka-flash-3:free", 32768),
-    ("Dolphin3.0 R1 Mistral 24B (free)", "cognitivecomputations/dolphin3.0-r1-mistral-24b:free", 32768),
-    ("Dolphin3.0 Mistral 24B (free)", "cognitivecomputations/dolphin3.0-mistral-24b:free", 32768),
-    ("Bytedance: UI-TARS 72B (free)", "bytedance-research/ui-tars-72b:free", 32768),
-    ("Qwerky 72b (free)", "featherless/qwerky-72b:free", 32768),
-    ("OlympicCoder 7B (free)", "open-r1/olympiccoder-7b:free", 32768),
-    ("OlympicCoder 32B (free)", "open-r1/olympiccoder-32b:free", 32768),
-    ("Rogue Rose 103B v0.2 (free)", "sophosympatheia/rogue-rose-103b-v0.2:free", 4096),
-    ("Toppy M 7B (free)", "undi95/toppy-m-7b:free", 4096),
-    ("Hugging Face: Zephyr 7B (free)", "huggingfaceh4/zephyr-7b-beta:free", 4096),
-    ("MythoMax 13B (free)", "gryphe/mythomax-l2-13b:free", 4096),
-    ("AllenAI: Molmo 7B D (free)", "allenai/molmo-7b-d:free", 4096),
+    # Smaller Models (<50B params)
+    {"category": "Smaller Models", "models": [
+        ("Gemma 3 1B", "google/gemma-3-1b-it:free", 32768),
+        ("Gemma 2 9B", "google/gemma-2-9b-it:free", 8192),
+        ("Mistral 7B Instruct", "mistralai/mistral-7b-instruct:free", 8192),
+        ("Qwen 2 7B Instruct", "qwen/qwen-2-7b-instruct:free", 8192),
+        ("Phi-3 Mini 128K Instruct", "microsoft/phi-3-mini-128k-instruct:free", 8192),
+        ("Phi-3 Medium 128K Instruct", "microsoft/phi-3-medium-128k-instruct:free", 8192),
+        ("OpenChat 3.5 7B", "openchat/openchat-7b:free", 8192),
+        ("Zephyr 7B", "huggingfaceh4/zephyr-7b-beta:free", 4096),
+        ("MythoMax 13B", "gryphe/mythomax-l2-13b:free", 4096),
+    ]},
 ]
 
+# Flatten model list for easy searching
+ALL_MODELS = []
+for category in MODELS:
+    for model in category["models"]:
+        ALL_MODELS.append(model)
+
 def format_to_message_dict(history):
     """Convert history to proper message format"""
     messages = []
@@ -122,35 +122,95 @@ def encode_image_to_base64(image_path):
         logger.error(f"Error encoding image: {str(e)}")
         return None
 
-def prepare_message_with_images(text, images):
-    """Prepare a message with text and images"""
-    if not images:
+def extract_text_from_file(file_path):
+    """Extract text from various file types"""
+    try:
+        file_extension = file_path.split('.')[-1].lower()
+        
+        if file_extension == 'pdf':
+            text = ""
+            with open(file_path, 'rb') as file:
+                pdf_reader = PyPDF2.PdfReader(file)
+                for page_num in range(len(pdf_reader.pages)):
+                    page = pdf_reader.pages[page_num]
+                    text += page.extract_text() + "\n\n"
+            return text
+        
+        elif file_extension == 'md':
+            with open(file_path, 'r', encoding='utf-8') as file:
+                md_text = file.read()
+                # You can convert markdown to plain text if needed
+                return md_text
+        
+        elif file_extension == 'txt':
+            with open(file_path, 'r', encoding='utf-8') as file:
+                return file.read()
+                
+        else:
+            return f"Unsupported file type: {file_extension}"
+            
+    except Exception as e:
+        logger.error(f"Error extracting text from file: {str(e)}")
+        return f"Error processing file: {str(e)}"
+
+def prepare_message_with_media(text, images=None, documents=None):
+    """Prepare a message with text, images, and document content"""
+    # If no media, return text only
+    if not images and not documents:
         return text
     
+    # Start with text content
+    if documents and len(documents) > 0:
+        # If there are documents, append their content to the text
+        document_texts = []
+        for doc in documents:
+            if doc is None:
+                continue
+            doc_text = extract_text_from_file(doc)
+            if doc_text:
+                document_texts.append(doc_text)
+        
+        # Add document content to text
+        if document_texts:
+            if not text:
+                text = "Please analyze these documents:"
+            else:
+                text = f"{text}\n\nDocument content:\n\n"
+            
+            text += "\n\n".join(document_texts)
+            
+        # If no images, return text only
+        if not images:
+            return text
+    
+    # If we have images, create a multimodal content array
     content = [{"type": "text", "text": text}]
     
-    for img in images:
-        if img is None:
-            continue
-        
-        encoded_image = encode_image_to_base64(img)
-        if encoded_image:
-            content.append({
-                "type": "image_url",
-                "image_url": {"url": encoded_image}
-            })
+    # Add images if any
+    if images:
+        for img in images:
+            if img is None:
+                continue
+            
+            encoded_image = encode_image_to_base64(img)
+            if encoded_image:
+                content.append({
+                    "type": "image_url",
+                    "image_url": {"url": encoded_image}
+                })
     
     return content
 
-def ask_ai(message, chatbot, model_choice, temperature, max_tokens, uploaded_files):
-    """Enhanced AI query function with file upload support and detailed logging"""
-    if not message.strip() and not uploaded_files:
+def ask_ai(message, chatbot, model_choice, temperature, max_tokens, top_p, frequency_penalty, 
+           presence_penalty, images, documents, reasoning_effort):
+    """Enhanced AI query function with comprehensive options"""
+    if not message.strip() and not images and not documents:
         return chatbot, ""
     
     # Get model ID and context size
     model_id = None
     context_size = 0
-    for name, model_id_value, ctx_size in MODELS:
+    for name, model_id_value, ctx_size in ALL_MODELS:
         if name == model_choice:
             model_id = model_id_value
             context_size = ctx_size
@@ -163,11 +223,8 @@ def ask_ai(message, chatbot, model_choice, temperature, max_tokens, uploaded_fil
     # Create messages from chatbot history
     messages = format_to_message_dict(chatbot)
     
-    # Prepare message with images if any
-    if uploaded_files:
-        content = prepare_message_with_images(message, uploaded_files)
-    else:
-        content = message
+    # Prepare message with images and documents if any
+    content = prepare_message_with_media(message, images, documents)
     
     # Add current message
     messages.append({"role": "user", "content": content})
@@ -175,16 +232,25 @@ def ask_ai(message, chatbot, model_choice, temperature, max_tokens, uploaded_fil
     # Call API
     try:
         logger.info(f"Sending request to model: {model_id}")
-        logger.info(f"Messages: {json.dumps(messages)}")
         
+        # Build the payload with all parameters
         payload = {
             "model": model_id,
             "messages": messages,
             "temperature": temperature,
-            "max_tokens": max_tokens
+            "max_tokens": max_tokens,
+            "top_p": top_p,
+            "frequency_penalty": frequency_penalty,
+            "presence_penalty": presence_penalty
         }
         
-        logger.info(f"Request payload: {json.dumps(payload)}")
+        # Add reasoning if selected
+        if reasoning_effort != "none":
+            payload["reasoning"] = {
+                "effort": reasoning_effort
+            }
+        
+        logger.info(f"Request payload: {json.dumps(payload, default=str)}")
         
         response = requests.post(
             "https://openrouter.ai/api/v1/chat/completions",
@@ -194,11 +260,10 @@ def ask_ai(message, chatbot, model_choice, temperature, max_tokens, uploaded_fil
                 "HTTP-Referer": "https://huggingface.co/spaces"
             },
             json=payload,
-            timeout=60
+            timeout=120  # Longer timeout for document processing
         )
         
         logger.info(f"Response status: {response.status_code}")
-        logger.info(f"Response headers: {response.headers}")
         
         response_text = response.text
         logger.info(f"Response body: {response_text}")
@@ -221,20 +286,68 @@ def ask_ai(message, chatbot, model_choice, temperature, max_tokens, uploaded_fil
     return chatbot, ""
 
 def clear_chat():
-    return [], "", [], 0.7, 1000
+    return [], "", [], [], 0.7, 1000, 0.8, 0.0, 0.0, "none"
+
+def filter_models(search_term):
+    """Filter models based on search term"""
+    if not search_term:
+        return gr.Dropdown.update(choices=[model[0] for model in ALL_MODELS], value=ALL_MODELS[0][0])
+    
+    filtered_models = [model[0] for model in ALL_MODELS if search_term.lower() in model[0].lower()]
+    
+    if filtered_models:
+        return gr.Dropdown.update(choices=filtered_models, value=filtered_models[0])
+    else:
+        return gr.Dropdown.update(choices=[model[0] for model in ALL_MODELS], value=ALL_MODELS[0][0])
+
+def get_model_info(model_name):
+    """Get model information by name"""
+    for model in ALL_MODELS:
+        if model[0] == model_name:
+            return model
+    return None
+
+def update_context_display(model_name):
+    """Update the context size display based on the selected model"""
+    model_info = get_model_info(model_name)
+    if model_info:
+        name, model_id, context_size = model_info
+        context_formatted = f"{context_size:,}"
+        return f"{context_formatted} tokens"
+    return "Unknown"
 
 # Create enhanced interface
-with gr.Blocks(css="footer {visibility: hidden}") as demo:
+with gr.Blocks(css="""
+    .context-size { 
+        font-size: 0.9em;
+        color: #666;
+        margin-left: 10px;
+    }
+    footer { display: none !important; }
+    .model-selection-row {
+        display: flex;
+        align-items: center;
+    }
+    .parameter-grid {
+        display: grid;
+        grid-template-columns: 1fr 1fr;
+        gap: 10px;
+    }
+""") as demo:
     gr.Markdown("""
     # Enhanced AI Chat
     
-    This interface allows you to chat with various free AI models from OpenRouter.
-    You can upload images for vision-capable models and adjust parameters.
+    Chat with various AI models from OpenRouter with support for images and documents.
     """)
     
     with gr.Row():
         with gr.Column(scale=2):
-            chatbot = gr.Chatbot(height=500, show_copy_button=True, show_label=False)
+            chatbot = gr.Chatbot(
+                height=500, 
+                show_copy_button=True, 
+                show_label=False,
+                avatar_images=(None, "https://upload.wikimedia.org/wikipedia/commons/0/04/ChatGPT_logo.svg")
+            )
             
             with gr.Row():
                 message = gr.Textbox(
@@ -251,93 +364,187 @@ with gr.Blocks(css="footer {visibility: hidden}") as demo:
                     clear_btn = gr.Button("Clear Chat", variant="secondary")
             
             with gr.Row():
-                uploaded_files = gr.Gallery(
-                    label="Uploaded Images", 
-                    show_label=True, 
-                    elem_id="gallery",
-                    columns=4, 
-                    height=150,
-                    visible=False
-                )
-            
-            with gr.Row():
-                upload_btn = gr.UploadButton(
-                    label="Upload Images (for vision models)",
-                    file_types=["image"],
-                    file_count="multiple"
-                )
+                # Image upload
+                with gr.Accordion("Upload Images (for vision models)", open=False):
+                    images = gr.Gallery(
+                        label="Uploaded Images", 
+                        show_label=True,
+                        columns=4, 
+                        height="auto",
+                        object_fit="contain"
+                    )
+                    
+                    image_upload_btn = gr.UploadButton(
+                        label="Upload Images",
+                        file_types=["image"],
+                        file_count="multiple"
+                    )
+                
+                # Document upload
+                with gr.Accordion("Upload Documents (PDF, MD, TXT)", open=False):
+                    documents = gr.File(
+                        label="Uploaded Documents",
+                        file_types=[".pdf", ".md", ".txt"], 
+                        file_count="multiple"
+                    )
         
         with gr.Column(scale=1):
             with gr.Group():
                 gr.Markdown("### Model Selection")
-                model_names = [name for name, _, _ in MODELS]
-                model_choice = gr.Radio(
-                    model_names,
-                    value=model_names[0],
-                    label="Choose a Model"
-                )
                 
-                with gr.Accordion("Model Context", open=False):
-                    context_info = gr.HTML(value="<p>Select a model to see its context window</p>")
-            
-            with gr.Accordion("Parameters", open=False):
-                temperature = gr.Slider(
-                    minimum=0.1, 
-                    maximum=2.0, 
-                    value=0.7, 
-                    step=0.1,
-                    label="Temperature"
-                )
+                with gr.Row(elem_classes="model-selection-row"):
+                    model_search = gr.Textbox(
+                        placeholder="Search models...",
+                        label="",
+                        show_label=False
+                    )
                 
-                max_tokens = gr.Slider(
-                    minimum=100, 
-                    maximum=4000, 
-                    value=1000, 
-                    step=100,
-                    label="Max Tokens"
-                )
+                with gr.Row(elem_classes="model-selection-row"):
+                    model_choice = gr.Dropdown(
+                        [model[0] for model in ALL_MODELS],
+                        value=ALL_MODELS[0][0],
+                        label="Model"
+                    )
+                    context_display = gr.Textbox(
+                        value=update_context_display(ALL_MODELS[0][0]),
+                        label="Context",
+                        interactive=False,
+                        elem_classes="context-size"
+                    )
+                
+                # Model category selection
+                with gr.Accordion("Browse by Category", open=False):
+                    model_categories = gr.Radio(
+                        [category["category"] for category in MODELS],
+                        label="Categories",
+                        value=MODELS[0]["category"]
+                    )
+                    
+                    category_models = gr.Radio(
+                        [model[0] for model in MODELS[0]["models"]],
+                        label="Models in Category"
+                    )
+            
+            with gr.Accordion("Generation Parameters", open=False):
+                with gr.Group(elem_classes="parameter-grid"):
+                    temperature = gr.Slider(
+                        minimum=0.0, 
+                        maximum=2.0, 
+                        value=0.7, 
+                        step=0.1,
+                        label="Temperature"
+                    )
+                    
+                    max_tokens = gr.Slider(
+                        minimum=100, 
+                        maximum=4000, 
+                        value=1000, 
+                        step=100,
+                        label="Max Tokens"
+                    )
+                    
+                    top_p = gr.Slider(
+                        minimum=0.1, 
+                        maximum=1.0, 
+                        value=0.8, 
+                        step=0.1,
+                        label="Top P"
+                    )
+                    
+                    frequency_penalty = gr.Slider(
+                        minimum=-2.0, 
+                        maximum=2.0, 
+                        value=0.0, 
+                        step=0.1,
+                        label="Frequency Penalty"
+                    )
+                    
+                    presence_penalty = gr.Slider(
+                        minimum=-2.0, 
+                        maximum=2.0, 
+                        value=0.0, 
+                        step=0.1,
+                        label="Presence Penalty"
+                    )
+                    
+                    reasoning_effort = gr.Radio(
+                        ["none", "low", "medium", "high"],
+                        value="none",
+                        label="Reasoning Effort"
+                    )
     
-    # Set up context window display
-    def update_context_info(model_name):
-        for name, _, ctx_size in MODELS:
-            if name == model_name:
-                return f"<p><b>Context window:</b> {ctx_size:,} tokens</p>"
-        return "<p>Model information not found</p>"
+    # Connect model search to dropdown filter
+    model_search.change(
+        fn=filter_models,
+        inputs=[model_search],
+        outputs=[model_choice]
+    )
     
+    # Update context display when model changes
     model_choice.change(
-        fn=update_context_info,
+        fn=update_context_display,
         inputs=[model_choice],
-        outputs=[context_info]
+        outputs=[context_display]
+    )
+    
+    # Update model list when category changes
+    def update_category_models(category):
+        for cat in MODELS:
+            if cat["category"] == category:
+                return gr.Radio.update(choices=[model[0] for model in cat["models"]], value=cat["models"][0][0])
+        return gr.Radio.update(choices=[], value=None)
+    
+    model_categories.change(
+        fn=update_category_models,
+        inputs=[model_categories],
+        outputs=[category_models]
+    )
+    
+    # Update main model choice when category model is selected
+    category_models.change(
+        fn=lambda x: x,
+        inputs=[category_models],
+        outputs=[model_choice]
     )
     
-    # Process uploaded files
-    def process_uploaded_files(files):
-        file_paths = [file.name for file in files]
-        return file_paths, gr.update(visible=True)
+    # Process uploaded images
+    def process_uploaded_images(files):
+        return [file.name for file in files]
     
-    upload_btn.upload(
-        fn=process_uploaded_files,
-        inputs=[upload_btn],
-        outputs=[uploaded_files, uploaded_files]
+    image_upload_btn.upload(
+        fn=process_uploaded_images,
+        inputs=[image_upload_btn],
+        outputs=[images]
     )
     
     # Set up events
     submit_btn.click(
         fn=ask_ai,
-        inputs=[message, chatbot, model_choice, temperature, max_tokens, uploaded_files],
+        inputs=[
+            message, chatbot, model_choice, temperature, max_tokens, 
+            top_p, frequency_penalty, presence_penalty, images, 
+            documents, reasoning_effort
+        ],
         outputs=[chatbot, message]
     )
     
     message.submit(
         fn=ask_ai,
-        inputs=[message, chatbot, model_choice, temperature, max_tokens, uploaded_files],
+        inputs=[
+            message, chatbot, model_choice, temperature, max_tokens, 
+            top_p, frequency_penalty, presence_penalty, images, 
+            documents, reasoning_effort
+        ],
         outputs=[chatbot, message]
     )
     
     clear_btn.click(
         fn=clear_chat,
         inputs=[],
-        outputs=[chatbot, message, uploaded_files, temperature, max_tokens]
+        outputs=[
+            chatbot, message, images, documents, temperature, 
+            max_tokens, top_p, frequency_penalty, presence_penalty, reasoning_effort
+        ]
     )
 
 # Launch directly with Gradio's built-in server