Unverified50% confidenceReleaseExact time
Red Note AI Lab launched a preview version of DOS 3 Note, a MoE model with 28 billion total parameters, only 1.6 billion activated parameters, supporting 512K ultra-long context and multimodality.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/8/2026
First Seen
Valid until: 12/7/2026
Sources
Related Entities
Related Claims
UnverifiedDuring the Think 2025 conference, IBM released a preview of the Granite 4 model, which adopts a Mixture of Experts (MoE) architecture with parameter sizes ranging from 3 billion to 20 billion, and the smallest version requires only 15 GB of memory to run a 120,000-token context window.62% similarUnverifiedOllama uses model quantization techniques (such as 4-bit and 8-bit quantization) to enable consumer-grade hardware (such as a MacBook with 16GB of memory) to run models with billions of parameters.62% similarUnverifiedLlama 3 8B模型（80亿参数）运行约需7GB显存62% similarUnverifiedThe 4-bit quantized GGUF format version of llama.cpp enables the Llama 3.1 70B model to run on an Apple M2 Ultra with 64GB unified memory at an inference speed of approximately 20–30 tokens/second.61% similarUnverifiedCombining MoE with quantization technology can make it possible for a laptop with 16 GB of memory to run MoE models with tens of billions of parameters.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/879002API
curl https://kongchang.com/api/v1/knowledge/claims/879002MCP
get_claim(id=879002)