Moonshot releases Kimi K3 weights with a 2.8T MoE and 1M-token context
teortaxesTex · x · 2026-07-28
Moonshot releases Kimi K3 weights and technical report.
- Kimi K3 is described as the company's most capable model so far.
- It uses a 2.8T MoE architecture with native visual understanding and a 1M-token context window.
- Moonshot says the new architecture delivers 2.5× more intelligence per unit of compute, not just a larger parameter count.
- The release also opens part of the stack behind the model, including high-performance attention kernels, an MoE communication library, and infrastructure for running agent environments at scale.
- Moonshot links the model weights, technical report, and tech blog with the announcement.
Related event: Moonshot AI Releases 2.8T Parameter Open-Weight Model Kimi K3(48 posts)→
More from Models
- Moonshot links the Kimi K3 report, weights and benchmark pages — ArtificialAnlys · 2026-07-28
- Moonshot releases Kimi K3 weights under a commercial-use-restricted license — ArtificialAnlys · 2026-07-28
- Chartography Benchmark: Zoom Tool Boosts Claude's Chart Accuracy to 44% — echen · 2026-07-28
- Testing K3 and Open Models: Reasoning Tokens Can Burn Entire Budgets — MaziyarPanahi · 2026-07-28
- Local Test: Nanbeige4.2-3B Lags Behind Qwen MoE in KV Cache Efficiency — TechTefa · 2026-07-28
- 35B Agentic Bakeoff: KAT-Coder Matches Qwen at Half the Token Cost — IvGranite · 2026-07-28