A flagship 70-billion-parameter model fine-tuned for chat completions, utilizing an optimized transformer architecture. Supports a context length of 4K tokens.