diff --git a/.coverage b/.coverage
index ee3e97a..c3b6e90 100644
Binary files a/.coverage and b/.coverage differ
diff --git a/gguf.sh b/gguf.sh
new file mode 100755
index 0000000..87f5058
--- /dev/null
+++ b/gguf.sh
@@ -0,0 +1,19 @@
+#!/usr/bin/env bash
+# Convert all safetensors to q8_0 GGUF format (skip if already exists)
+
+set -euo pipefail
+
+for sf in *.safetensors; do
+    [[ -f "$sf" ]] || continue
+
+    base="${sf%.safetensors}"
+    gguf="${base}-q8_0.gguf"
+
+    if [[ -f "$gguf" ]]; then
+        echo "Skip: $gguf exists"
+        continue
+    fi
+
+    echo "Converting: $sf -> $gguf"
+    sd -M convert -m "$sf" -o "$gguf" --type q8_0
+done
diff --git a/tensors/cli.py b/tensors/cli.py
index d0b3b1c..5b12c12 100644
--- a/tensors/cli.py
+++ b/tensors/cli.py
@@ -48,6 +48,18 @@ from tensors.safetensor import compute_sha256, get_base_name, read_safetensor_me
 # Key masking threshold
 MIN_KEY_LENGTH_FOR_MASKING = 8
 
+# Size threshold for GB display
+_MB_PER_GB = 1024
+
+
+def _format_size_mb(size_mb: float | None) -> str:
+    """Format size in MB to human-readable string."""
+    if not size_mb:
+        return ""
+    if size_mb >= _MB_PER_GB:
+        return f"{size_mb / _MB_PER_GB:.1f} GB"
+    return f"{size_mb:.0f} MB"
+
 
 def _version_callback(value: bool) -> None:
     if value:
@@ -1016,16 +1028,25 @@ def models_list(
         return
 
     table = Table(title="Available Models", show_header=True, header_style="bold magenta")
-    table.add_column("Status", style="dim", width=3)
+    table.add_column("ID", style="dim", width=8)
     table.add_column("Name", style="cyan")
-    table.add_column("Path", style="dim")
+    table.add_column("File", style="white")
+    table.add_column("Size", style="green", justify="right")
 
     for model in models:
         path = model.get("path", "")
         name = model.get("name", Path(path).stem if path else "")
         is_active = active in {path, name}
-        status = "[green]✓[/green]" if is_active else ""
-        table.add_row(status, name, path)
+
+        civitai_id = model.get("civitai_model_id")
+        id_str = str(civitai_id) if civitai_id else ""
+        display_name = model.get("display_name", name)
+        if is_active:
+            display_name = f"[green]✓[/green] {display_name}"
+        filename = model.get("filename", Path(path).name if path else "")
+        size_str = _format_size_mb(model.get("size_mb"))
+
+        table.add_row(id_str, display_name, filename, size_str)
 
     console.print(table)
 
@@ -1091,13 +1112,22 @@ def models_loras(
         return
 
     table = Table(title="Available LoRAs", show_header=True, header_style="bold magenta")
+    table.add_column("ID", style="dim", width=8)
     table.add_column("Name", style="cyan")
-    table.add_column("Path", style="dim")
+    table.add_column("File", style="white")
+    table.add_column("Size", style="green", justify="right")
 
     for lora in loras:
         path = lora.get("path", "")
         name = lora.get("name", Path(path).stem if path else "")
-        table.add_row(name, path)
+
+        civitai_id = lora.get("civitai_model_id")
+        id_str = str(civitai_id) if civitai_id else ""
+        display_name = lora.get("display_name", name)
+        filename = lora.get("filename", Path(path).name if path else "")
+        size_str = _format_size_mb(lora.get("size_mb"))
+
+        table.add_row(id_str, display_name, filename, size_str)
 
     console.print(table)
 
diff --git a/tensors/server/generate_routes.py b/tensors/server/generate_routes.py
index 2a52a87..75bc8ec 100644
--- a/tensors/server/generate_routes.py
+++ b/tensors/server/generate_routes.py
@@ -5,6 +5,7 @@ from __future__ import annotations
 import base64
 import logging
 import time
+from pathlib import Path
 from typing import Any
 
 import httpx
@@ -55,8 +56,16 @@ class GenerateRequest(PydanticBaseModel):
 
 def _build_sd_request(req: GenerateRequest) -> dict[str, Any]:
     """Build request body for sd-server."""
+    prompt = req.prompt
+
+    # sd-server expects LoRA in prompt as <lora:name:weight> syntax
+    if req.lora:
+        lora_name = Path(req.lora.path).stem
+        lora_tag = f"<lora:{lora_name}:{req.lora.multiplier}>"
+        prompt = f"{prompt} {lora_tag}"
+
     body: dict[str, Any] = {
-        "prompt": req.prompt,
+        "prompt": prompt,
         "negative_prompt": req.negative_prompt,
         "width": req.width,
         "height": req.height,
@@ -69,9 +78,6 @@ def _build_sd_request(req: GenerateRequest) -> dict[str, Any]:
         body["sampler_name"] = req.sampler_name
     if req.scheduler:
         body["scheduler"] = req.scheduler
-    # sd-server expects LoRA as JSON array, not in prompt
-    if req.lora:
-        body["lora"] = [{"path": req.lora.path, "multiplier": req.lora.multiplier}]
     return body