Moonshot releases Kimi K3 with 2.8T MoE, native vision and 1M-token context
drdanielbender · x · 2026-07-28
Moonshot releases Kimi K3, a 2.8T MoE model with 1M-token context and native vision
Moonshot says Kimi K3 is its most capable model so far. The release includes:
- 2.8T MoE architecture
- Native visual understanding
- 1 million token context window
- A claim of 2.5x intelligence per unit of compute through a new architecture
Alongside the weights, Moonshot is also opening up parts of the stack behind the model, including high-performance attention kernels, an MoE communication library, and infrastructure for running agent environments at scale.
The post shares the model weights, technical report, and tech blog, positioning Kimi K3 as both a frontier model release and a deeper stack disclosure.
Related event: Moonshot AI Releases Open-Weight Kimi K3 Model(37 posts)→
More from Infra
- Gemma 4 is benchmarked locally on a 48GB Mac with MLX, llama.cpp and Java 25 — rseroter · 2026-07-28
- Bittensor subnet expansion is pitched as a cheaper AI infrastructure path for companies — markjeffrey · 2026-07-28
- Project Orion is training a 16B model live across three continents on heterogeneous compute — markjeffrey · 2026-07-28
- Modular handbook maps the hidden costs of LLM inference, from KV cache to prefill/decode splits — udmrzn · 2026-07-28
- Kimi K3 reaches Merge Gateway with U.S. inference providers and ZDR terms — shensi · 2026-07-28
- Compute, not algorithms, is the real moat in frontier AI — GavinSBaker · 2026-07-28