Unverified50% confidenceBenchmarkExact time
动态GGUF的UD-IQ2_XXS档位将原始BF16模型体积从54.7GB压缩到约9GB，体积压缩率83.5%，准确率保留82.5%
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/24/2026
First Seen
Valid until: 11/22/2026
Sources
Related Entities
Related Claims
UnverifiedFP8 quantization can compress the weights of a 7B model from approximately 14GB (FP16) to approximately 7GB.76% similarUnverifiedThe original QW3.6 27B model file size is approximately 54 GB; Prismo compresses it to under 4 GB through mathematical compression techniques, a compression ratio of more than 13×, and claims this does not affect model performance.76% similarUnverifiedGGUF 量化格式可将FP16或BF16精度模型压缩至4-bit乃至2-bit精度75% similarUnverifiedA 70B model can be compressed to approximately 35GB at INT4 precision.75% similarUnverified一个1T参数模型以BF16精度存储需要约2TB显存，INT4量化压缩也需500GB以上75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/795628API
curl https://kongchang.com/api/v1/knowledge/claims/795628MCP
get_claim(id=795628)