Skip to main content

Models

Kimi K3 for Engineers: A 2.8T-Parameter Open-Weight Model You Can Actually Self-Host

An engineer's briefing on Moonshot AI's Kimi K3 — the 2.8T/104B-active MoE with Kimi Delta Attention and LatentMoE, why it's open-weight and not OSI open-source, real hardware requirements, and how to evaluate it against closed frontier models.

AI Playbook2026-07-2910 min readadvanced

Executive takeaway

An engineer's briefing on Moonshot AI's Kimi K3 — the 2.8T/104B-active MoE with Kimi Delta Attention and LatentMoE, why it's open-weight and not OSI open-source, real hardware requirements, and how to evaluate it against closed frontier models.

In this briefing

  1. Key ideas
  2. Practical implications
  3. What to do next

Why it matters

Use this briefing to decide whether to deep-dive the full article for your current delivery problem.

AI EngineerConsultantStudent

Turn insight into action

Read the full analysis, then continue into a playbook or framework.