Moonshot AI Releases Kimi K3: 2.8 Trillion Parameter Open MoE Model
Decision Brief
What changedMoonshot AI launched Kimi K3 on July 16, 2026, a 2.8 trillion parameter open MoE model.
Why it mattersWith 2.8T parameters but only 16 of 896 experts activated, inference cost is far lower than full-parameter models, ideal for teams deploying large open-source models.
Who should careOpen-source model users
Affected stackKimi
Source confidenceMedium · Reliable media or first-hand reporting
Moonshot AI released Kimi K3 on July 16, 2026, a 2.8 trillion parameter open MoE model. It uses Kimi Delta Attention and Attention Residuals, activating 16 out of 896 experts, with a 1M context length. For developers and research teams needing long-context processing, Kimi K3's 1M context window and sparse activation design reduce computation while maintaining performance. Companies adopting this model can save inference resources, especially for document analysis and codebase understanding tasks.
Summary basis: official / RSS sourceCompiled from the source scope noted above; the original remains authoritative.
Sources
- MarkTechPost
Fast research-paper and ML tooling summaries, useful for infra and agent updates.
- MarkTechPost
留言
登入后即可留言，和其他 builder 交换实测心得。
还没有留言，抢头香。