Changelog

v2.0.30 (2025-11-01)

Features

• Improved Responsive Layout with collapsible sidebar
• Auto-reload when config files change
• Cancel Button to cancel pending requests
• Auto-focus returns to textarea after request completes
• Better UX for model and system prompt selectors
  • Clicking outside selector collapses it
  • Clicking selected item no longer deselects it
• Environment Variable Support for VERBOSE=1 to enable verbose mode (useful in Docker)

v2.0.28 (2025-10-31)

Major Features

• Dark Mode support
• Drag & Drop files in message prompt
• Copy & Paste files in message prompt
• GitHub OAuth authentication
  • Optional restrict access to specified users
• Docker Support
  • Official Docker images
  • Docker Compose configuration

v2.0.24

Analytics & Metrics

• Cost Tracking

  • Model selector displays input/output cost per 1M tokens
  • Thread-level cost and token metrics in sidebar
  • Per-message token breakdown
  • Thread summaries with total cost, tokens, requests, and response time

• Analytics Pages

  • Cost Analytics: Daily cost breakdown by month with per-model/provider details
  • Token Analytics: Daily token usage by month with input/output breakdown
  • Activity Log: Detailed request history with model, provider, tokens, cost, and performance metrics

• Provider Reliability

  • --check command to test provider status and response times
  • GitHub Action to publish provider test results

UI Improvements

• Edit & Redo: Modify or re-run existing prompts
• Metrics Display: Token counts and costs visible throughout UI
• Export/Import: Backup and restore analytics data (hold ALT while clicking Export)

Earlier Versions

Core Features

• Ultra-Lightweight Architecture

  • Single file implementation
  • Single aiohttp dependency
  • Zero dependencies for ComfyUI integration

• Multi-Provider Gateway

  • Support for 160+ models
  • Automatic failover
  • Intelligent routing
  • Cost optimization

• Multi-Modal Support

  • Text generation
  • Vision (images)
  • Audio processing
  • File attachments (PDFs)

• Flexible Deployment

  • CLI tool
  • HTTP server (OpenAI-compatible)
  • Python module
  • ComfyUI node

• Web UI

  • ChatGPT-like interface
  • Offline operation
  • Local storage (IndexedDB)
  • Markdown rendering
  • Syntax highlighting
  • 200+ system prompts
  • Search history
  • Import/export

• Provider Support

  • OpenRouter (free and paid)
  • Groq
  • Google (free and paid)
  • Codestral
  • Ollama (local)
  • OpenAI
  • Anthropic
  • Grok (X.AI)
  • Qwen
  • Z.ai
  • Mistral

• Configuration Management

  • JSON-based configuration
  • Environment variable support
  • Easy provider enable/disable
  • Custom model mappings
  • Pricing configuration

Upgrade Instructions

From pip

pip install llms-py --upgrade

From Docker

# Pull latest image
docker pull ghcr.io/servicestack/llms:latest

# Restart container
docker compose restart

Configuration Migration

Configuration files are automatically migrated when you upgrade. Your existing settings in ~/.llms/llms.json and ~/.llms/ui.json will be preserved.

If you encounter issues, you can:

1. Backup current config:

cp ~/.llms/llms.json ~/.llms/llms.json.backup
cp ~/.llms/ui.json ~/.llms/ui.json.backup

2. Reset to defaults:

rm -rf ~/.llms
llms --init

3. Restore your customizations from the backup files

Breaking Changes

v2.0.28

• GitHub OAuth configuration moved to ui.json
• New auth section in configuration

v2.0.24

• Analytics data structure changed
• Old analytics data may need to be cleared

Roadmap

Planned features for future releases:

• Streaming Support: Real-time response streaming in UI
• More Providers: Additional LLM provider integrations
• Advanced Analytics: More detailed usage insights
• Team Features: Multi-user support
• API Enhancements: Additional API endpoints
• Plugin System: Extensibility for custom providers

Links

• GitHub Repository: github.com/ServiceStack/llms
• PyPI Package: pypi.org/project/llms-py
• Release Notes: github.com/ServiceStack/llms/releases
• Issues: github.com/ServiceStack/llms/issues

Contributing

We welcome contributions! See the GitHub repository for:

• Bug reports
• Feature requests
• Pull requests
• Documentation improvements