Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.2.0] - 2025-05-10

Added

• Google Gemini Support - Full integration with Google Gemini models
  • Gemini Pro, Gemini Pro Vision, and Gemini Ultra support
  • Streaming responses support
  • Vision/multimodal capabilities
  • Safety settings configuration
  • Generation parameters (temperature, top-k, top-p)
• Gemini provider implementation with comprehensive error handling
• Gemini-specific test suite
• Gemini usage examples

Changed

• Updated provider support matrix to show Gemini as ✅ Supported
• Enhanced documentation with Gemini configuration and examples

0.1.0 - 2025-05-10

Added

• Initial release of LiteLLM Clojure
• Unified API for multiple LLM providers
• Support for OpenAI (GPT-3.5-Turbo, GPT-4, GPT-4o)
• Support for Anthropic (Claude 3 Opus, Sonnet, Haiku, Claude 2.x)
• Support for OpenRouter (access to multiple providers)
• Support for Ollama (local models)
• Async operation support with proper thread pool management
• Streaming response support for OpenAI and Anthropic
• Function calling support for OpenAI models
• Provider abstraction layer for easy extension
• Health monitoring and system checks
• Cost tracking and token estimation
• Request/response schema validation with Malli
• Comprehensive error handling
• Thread pool management with Claypoole
• Configuration system with Aero support
• Built-in caching support

Core API

• create-system - Create and configure LiteLLM system
• shutdown-system! - Gracefully shutdown system
• completion - Main completion API
• make-request - Direct request API
• health-check - System health monitoring
• system-info - Get system information

Provider Support Matrix

• ✅ OpenAI - Full support with streaming and function calling
• ✅ Anthropic - Full support with streaming
• ✅ OpenRouter - Full support with multiple model access
• ✅ Ollama - Basic support for local models
• 🔄 Azure OpenAI - Planned
• 🔄 Google (Gemini) - Planned
• 🔄 Cohere - Planned
• 🔄 Mistral - Planned

Known Issues

• Test suite needs refinement for configuration validation
• Provider name extraction logic needs improvement for some edge cases
• Integration tests require API keys and are tagged separately
• Streaming API implementation is partial
• Some advanced OpenAI features not yet supported

Documentation

• Comprehensive README with installation and usage examples
• Provider-specific configuration guides
• API reference examples
• Quick start guide

Dependencies

• Clojure 1.11.1
• Hato 0.9.0 (HTTP client)
• Cheshire 5.12.0 (JSON)
• Claypoole 1.1.4 (Thread pools)
• Malli 0.13.0 (Schema validation)
• Core.async 1.6.681 (Async operations)

Build & Distribution

• Maven/Clojars compatible build system
• tools.build integration
• Group ID: tech.unravel
• Artifact ID: litellm-clj