Unverified50% confidenceBenchmarkExact time
Mixtral 8x7B was the first open-source MoE model to attract widespread industry attention, using 8 7B experts (46.7B total parameters, 12.9B activated) to achieve performance close to Llama 2 70B.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/5/2026
First Seen
Valid until: 12/4/2026
Sources
Related Entities
Related Claims
UnverifiedTypical representatives of the MoE architecture include Mixtral 8x7B and DeepSeek-V2.69% similarUnverifiedZonos 2 采用 MoE 混合专家架构，总参数量达 8B，激活参数仅为 900M65% similarUnverifiedMixtral 8x7B每次前向传播仅激活2个专家63% similarUnverifiedMixtral 8x7B, released by Mistral AI, is the first mainstream open-weight MoE model, with 46.7B total parameters and 12.9B active parameters.61% similarUnverifiedWith the maturation of local inference frameworks such as llama.cpp and Ollama, and advances in quantization techniques (such as GGUF and AWQ), running models with 7B–13B parameter scales on consumer-grade hardware has become possible.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/858932API
curl https://kongchang.com/api/v1/knowledge/claims/858932MCP
get_claim(id=858932)