Your browser does not seem to support JavaScript. As a result, your viewing experience will be diminished, and you have been placed in read-only mode.
Please download a browser that supports JavaScript, or enable it if it's disabled (i.e. NoScript).
我和你的机器配置一样,也测试了这个模型,接入了Cindy,500k上下文,基本能在1我目前也测试了这个模型并接入了Cindy,500k上下文,基本能在20tokens每秒。之前也用了一段时间的MLX-serve和omlx来推Qwen3.8 27b,从响应速度,工具调用,任务分析体感上相比这个方案都要更好一些。