Event ·Multi-source

阿里巴巴发布Qwen3.8-Flash-Next，目标‘极致成本效率’

阿里巴巴Qwen团队发布Qwen3.8-Flash-Next，混合专家模型，每token激活125亿参数中的6个。训练成本降低至九分之一，在编码和办公基准测试中击败DeepSeek-V4-Flash和Claude Opus 4.6，对OpenAI和Anthropic构成更多价格压力。

Current read

想了解阿里巴巴如何以更低成本打造强大模型？Qwen3.8-Flash-Next值得一试，它比DeepSeek-V4-Flash和Claude Opus 4.6更高效，成本更低。

11 sources73° AI heatLast updated 2026/8/26 14:40:09

Evidence

11 sources

Primary sourceDecoder

2026/8/26 14:40:09View source

Related source 1lmarena.ai

2026/8/26 15:24:04View source

Related source 2阿里通义 Qwen

2026/8/26 13:17:23View source

Related source 3marktechpost

2026/8/26 15:20:00View source

Related source 4NVIDIA AI

2026/8/26 17:17:05View source

Related source 5orange.ai

2026/8/25 13:27:25View source

Related source 6IT之家

2026/8/25 14:51:37View source

Related source 7arXiv: DeepSeek

2026/8/26 10:05:33View source

Related source 8SuperTechFans

2026/8/27 00:44:28View source

Related source 9AI Will

2026/8/26 08:51:35View source

Related source 10Geek

2026/8/26 14:14:10View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story