Unverified50% confidenceBenchmarkExact time
For ModelFit on Kimi K3 (unsloth repository), the UD-Q1_0 quantization has a measured file size of 466.4 GB and requires 561.1 GB of VRAM/RAM at an 8K context; UD-Q4_K_XL has a measured file size of 1.51 TB and requires about 1.6 TB; F16 is about 3 TB.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/17/2026
First Seen
Valid until: 10/31/2026
Sources
Related Entities
Related Claims
UnverifiedFor Kimi K3, the VRAM requirement answers given by five popular pages were 594 GB, 1.51 TB, 1.68 TB, and "can't run it", and for the Q4 quantized requirement there were two claims, "about 800 GB" and 1.2 TB.78% similarUnverifiedConverting the 8-bit Ngram table to 4-bit can save approximately 100 GB, reducing the model size from approximately 475 GB to approximately 385 GB.75% similarUnverifiedFor a 70B model using GQA (n_layers=80, n_kv_heads=8, head_dim=128, FP16 cache), the KV Cache size per token is about 320 KB.75% similarUnverifiedKimi K2在FP16精度下仅模型权重就需要数百GB显存，INT4量化可将显存需求压缩至约四分之一75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1054742API
curl https://kongchang.com/api/v1/knowledge/claims/1054742MCP
get_claim(id=1054742)