Moonshot releases Kimi K3 weights, a 2.8T MoE model with 1M-token context
KyeGomezB · x · 2026-07-28
Moonshot AI has released the weights and technical report for Kimi K3, its most capable model so far: a 2.8T MoE system with native visual understanding and a 1M-token context window.
The company says K3 uses a new architecture that delivers 2.5x more intelligence per unit of compute, not just more parameters. Alongside the model, Moonshot is also opening up more of the stack behind it, including:
- high-performance attention kernels
- an MoE communication library
- infrastructure for running agent environments at scale
A third-party Swarms tutorial also shows how to connect Kimi K3, build a first AI agent, and create a multi-agent GroupChat in a few lines of code.
Related event: Moonshot AI Releases 2.8T Open-Weight Model Kimi K3(121 posts)→
More from coding & agent
- Qwen 3.6 27B agent gets much smarter after KV cache quantization change — Jordanthecomeback · 2026-07-28
- Using K3 with Codex Desktop was a “big mistake,” says developer — HamelHusain · 2026-07-28
- Local dual-agent setup catches an AI trying to swap SQLite for Postgres — PrajwalTomar_ · 2026-07-28
- Microsoft Research: LLMs Fail to Track Evolving User Intent in Multi-Turn Conversations — alan_ritter · 2026-07-28
- Stop Building Everything: Why AI Startups Should Focus on Component-Level Breakouts — zeeg · 2026-07-28
- Kimi K3 tokenizer optimization cuts first-token latency by about 325 ms — philipkiely · 2026-07-28