v1.1.5

v1.1.5

• New ps Command - Show running models and their status
  • Display all models currently loaded in memory
  • Works with both local and remote Ollama servers
  • Shows NAME, ID, SIZE, VRAM USAGE, CONTEXT, UNTIL in formatted table
  • Same output format as run/chat command preload display
• Chat Model Preloading - Models are now automatically loaded into memory before first chat input
  • Sends empty chat request to preload model
  • Displays loaded model status table after preload
  • Shows model name, ID (shortened digest), size, VRAM usage, context length, and expiration time
• Process Status Display - New formatted table showing all loaded models via /api/ps
  • NAME: Model name with truncation for long names
  • ID: First 12 characters of model digest (Docker-style)
  • SIZE: Disk size combined with parameter count (e.g., "4.54 GB (8.0B)")
  • VRAM USAGE: Memory allocated in VRAM
  • CONTEXT: Context window size (e.g., 4096)
  • UNTIL: Human-readable expiration time (e.g., "2 minutes from now", "About a minute from now")
• Improved Chat Performance - Fixed streaming response buffering for remote servers
  • Uses HttpCompletionOption.ResponseHeadersRead for immediate streaming
  • Remote chat now responds as fast as local chat
  • No more delays waiting for full response buffering