Trinity Mini

A 26B sparse mixture-of-experts language model with 3B parameters active per token and 131k token context window, designed for efficient reasoning over extended documents.

Strengths

The sparse routing mechanism keeps computational overhead low while maintaining performance on reasoning tasks, and the large context window enables processing of lengthy inputs without truncation.

Best for

Applications requiring long-document analysis, summarization, or reasoning where inference speed and efficiency matter more than maximum reasoning capability.

Limitations

With only 3B active parameters, it has less reasoning depth than larger dense models, and sparse MoE architectures can show uneven performance across diverse task types.