Unverified60% confidenceFactExact time
The DeepSeek distilled versions range in parameter count from 1.5B to 70B.
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
9/10/2026
First Seen
Valid until: 12/9/2026
Sources
Related Entities
Related Claims
UnverifiedDeepSeek V3's activated parameters are compressed to 3 to 5 billion, with a total parameter count of 671B.76% similarUnverifiedSeries such as Qwen and DeepSeek all offer distilled or small-size versions such as 1.5B, 7B and 8B.75% similarUnverifiedAccording to community-cited information, DeepSeek-V4.1-Flash has 552B backbone parameters, 196B Engram parameters, and safetensors weight files totaling approximately 350 GB.72% similarUnverifiedDeepSeek published a sparse prefill figure of 1350 TFLOPS (absolute value) for the NVIDIA B200, which uses a different metric from the 95% peak utilization figure cited for Ascend.72% similarUnverifiedIn the DeepSeek V1 era, each incremental token required nearly 400 KB of KV cache, which dropped to 48 KB in V3.2.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/886796API
curl https://kongchang.com/api/v1/knowledge/claims/886796MCP
get_claim(id=886796)