Cartik/Bastiai-1-instruct

TEXT GENERATIONConcurrency Cost:1Model Size:0.5BQuant:BF16Ctx Length:32kPublished:Apr 29, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

Cartik/Bastiai-1-instruct is a 0.5 billion parameter instruction-tuned causal language model based on Qwen2.5-0.5B-Instruct, developed by Cartik. It supports both Russian and English languages, making it suitable for multilingual applications. With a context length of 32768 tokens, this model is optimized for tasks requiring processing of longer inputs in both supported languages.

Loading preview...

Cartik/Bastiai-1-instruct Overview

Cartik/Bastiai-1-instruct is a compact yet capable 0.5 billion parameter instruction-tuned language model. Built upon the Qwen2.5-0.5B-Instruct architecture, this model is specifically designed to handle tasks in both Russian (ru) and English (en), making it a strong candidate for bilingual applications.

Key Capabilities

  • Bilingual Support: Proficient in processing and generating text in both Russian and English.
  • Instruction Following: Fine-tuned to understand and execute instructions effectively.
  • Extended Context: Features a substantial context window of 32768 tokens, allowing it to process and reason over longer documents and conversations.

Good for

  • Applications requiring efficient processing of text in both Russian and English.
  • Tasks that benefit from a large context window, such as summarization of long articles or complex dialogue management.
  • Deployment in resource-constrained environments where a smaller model size is advantageous without sacrificing multilingual capability.