Qwen3.8-27B: Free AI Model Outperforms Claude Opus

A new AI model called Qwen3.8-27B has been released for free. This locally runnable version, from Alibaba's Qwen team, has 27 billion parameters. What's interesting is it surpasses Claude Opus 4.6 Max in some performance benchmarks. Its weights are under the Apache License 2.0, allowing commercial use with certain conditions. This model can be installed on any PC. The Qwen team describes it as a compact and easily deployable model. It uses a Dense model approach, utilizing the entire model during inference. An FP8 quantized version is also available to reduce size and computational load. The bottom line is this release makes powerful AI more accessible for individuals and smaller development environments.