←Back to NewsAI News/LlmsnewsLlmsOpen SourceAlibaba's Qwen3.8-Flash-Next Hits 10x Long-Context Speed at 1/9 Training CostAlibaba open-weights a 125B multimodal MoE with just 6B active parameters, previewing the attention overhaul coming in Qwen4.SourceQwenPublishedAug 26, 2026, 12:34 PMAuthorAlphaSignal NewsroomRead1 min readAlibaba open-weights a 125B multimodal MoE with just 6B active parameters, previewing the attention overhaul coming in Qwen4.Reporting is indexed from AlphaSignal. Rights remain with the original publisher and cited sources.Read original report ↗Next readsAlphaSignal · repoOpenJev Runs Agent Decisions 5x Faster by Skipping Token GenerationArtificial Analysis · newsAnt Group Releases Ling-3.0-flash-Fin to Self-Host Financial Research at 124B ParametersAlphaSignal · modelChina Telecom Releases Xing4.0, an Open 29B Coding Agent Built on Huawei Chips
