[KongchangAI]
Unverified50% confidenceFactExact time

The model size of V4-Flash-Vision is approximately 168 GB (FP8 weights), while the original V4-Flash is 160 GB, meaning the multimodal vision capability adds only about 8 GB of weights.

1
Sources
50%
Confidence
Long-term
Relevance
9/11/2026
First Seen

Sources

DeepSeek V4-Flash-Vision Open-Source Hands-On: An 8GB Vision Module Gives the Model Eyes

Related Entities

V4-Flash-VisionmodelV4-Flashmodel

Related Claims

UnverifiedV4 Flash's cache size is only 7% of that of V3.2.75% similarUnverifiedAfter compression, the DeepSeek V4 Flash model size was reduced from nearly 500 GB to 80.8 GB.73% similar

Cite This Claim

Stable URI
https://kongchang.com/claim/906104
API
curl https://kongchang.com/api/v1/knowledge/claims/906104
MCP
get_claim(id=906104)