Qwen 3 Model Series Announced with Qwen3.8-Max and Open Weights
The upcoming Qwen 3 model series, featuring the 2.4T parameter Qwen3.8-Max and the Qwen3.8-27B, has been announced with open weights releasing next week. Qwen3.8-Max matches the performance of Kimi K3 and DeepSeek V4 Flash, while showing superior capabilities in coding and software tasks. The release of open weights for massive models like Qwen3.8-Max (2.4T) provides the open-source AI community with state-of-the-art capabilities that rival top-tier proprietary and open-weight models. This intensifies competition in the LLM landscape, particularly for complex coding and software engineering applications. API pricing for Qwen3.8-Max is set at $2.0 per million input tokens, $6.0 per million output tokens, and $0.25 per million tokens for implicit caching. The model series includes both the massive 2.4T parameter Max version and a smaller 27B parameter version.
## BACKGROUND
Qwen is a prominent series of large language models developed by Alibaba, known for strong benchmark performance. Kimi K3 is a 2.8T parameter open-weight multimodal reasoning model from Moonshot AI, while implicit caching is a technique that automatically optimizes performance and costs by reusing previously processed context prefixes.