XHToken has released two new language models, Spark-X2.5-4B and Spark-X2.5-1.7B, available on Hugging Face. These models are designed for efficiency and accessibility, offering strong performance across various tasks like conversation, coding, and agentic workflows. Key features include a hybrid attention architecture enabling a native 1M-token context window and support for over 200 languages, positioning them as leading open-source options for their size. AI
IMPACT These models offer efficient, long-context capabilities, potentially lowering the barrier for advanced AI applications.
RANK_REASON New model release from an AI lab (XHToken) with technical details and benchmark claims. [lever_c_demoted from frontier_release: ic=2 ai=1.0]
Read on Hugging Face Trending Models →
- Docker
- Google Colab
- Hugging Face
- Kaggle
- OpenAI
- SGLang
- transformers
- vLLM
- XHToken
- XHToken/Spark-X2.5-4B
- Spark-X2.5
- XHToken/Spark-X2.5-1.7B
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →