ASTRAI Labs

Building efficient conversational AI and language models.

ASTRAI Labs develops language models focused on useful, capable and efficient chatbot experiences.

Focus areas

• Conversational AI
• Large language models
• Reasoning and instruction following
• General knowledge
• Model distillation
• Preference optimization
• Efficient training and inference
• Multilingual AI

Models

Models and releases will be published here.

Research direction

We explore practical methods for improving compact language models through pre-training, supervised fine-tuning, preference optimization, distillation and reasoning-focused training.