Qwen3.5 122B (a10b)

Qwen3.5 122B A10B Instruct is part of the Qwen3.5 model family from Alibaba, featuring a Mixture-of-Experts architecture with 122 billion total parameters and 10 billion active per token. This highly efficient design provides strong quality with minimal inference cost.

The model represents an efficiency-focused approach in the Qwen lineup, offering competitive performance on reasoning, coding, and multilingual tasks while activating only a small fraction of its parameters. It's designed for high-throughput production deployments where cost efficiency is important.

Qwen3.5 122B A10B is an excellent choice for applications requiring good quality at very low inference cost, offering one of the best quality-per-dollar ratios in the Qwen model family.