Moonshot’s Kimi K3 now runs locally at 594 GB with 78.9% top-1 accuracy
danielhanchen · x · 2026-07-29
Moonshot’s Kimi K3 is now runnable locally through Unsloth’s quantized builds.
- The model is described as a 2.8T-parameter open-weight model with 104B active parameters, native vision, and a 1M-token context window.
- Unsloth reports a dynamic 1-bit build at 594 GB, down from 1.56 TB, while retaining about 78.9% top-1 accuracy.
- A dynamic 2-bit build reaches about 90% accuracy at 861.3 GB, and the model can run on setups such as a Mac Studio with 128 GB RAM or an NVIDIA DGX Station-class machine.
- The guide also includes hardware requirements, quantization comparisons, and notes on running the model via Unsloth Studio or llama.cpp.
Related event: Unsloth Releases Kimi K3 Quantized Weights, 1-bit Compresses to 594GB(6 posts)→
More from coding & agent
- Opus 5 Shreds Repo While Dev Scrolls X: A Coding Agent Meme — verrsane · 2026-07-30
- Scale AI NYC Meetup: Advancements in Benchmarking Coding Agents — jyangballin · 2026-07-30
- OpenClaw Exposes Critical RCE Flaw, 50k+ Nodes Compromised — ericelliott_ · 2026-07-30
- Dev tests two adversarial agents, they just keep catching each other cheating — Vjeux · 2026-07-30
- 5 AI Engineering Trends: Coding Agents Set to Replace IDEs — The AI Daily Brief · 2026-07-30
- CrowdStrike Report Reveals Hidden Vulnerabilities in AI-Generated Code — RexDouglass · 2026-07-30