Unverified50% confidenceFactExact time
DeepSeek V4 Flash是主打高吞吐的云端模型，推理速度可达约每秒140 tokens
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/11/2026
First Seen
Valid until: 12/10/2026
Sources
Related Entities
Related Claims
Unverified在双Spark系统上，DeepSeek V4 Flash（FP8量化）峰值约每秒53个Token79% similarVerifiedDeepSeek V4 Flash在OpenModel平台的速率限制为每分钟10次请求（10 RPM）和每分钟100K tokens（100K TPM）78% similarUnverified在Artificial Analysis的成本基准中，DeepSeek V4 Flash位于成本最低区间，速度约110 tokens/s76% similarUnverifiedDeepSeek-V4-Flash-Vision-Exp采用MoE架构，总参数约305B，每个Token仅激活约13B，激活率约4.6%74% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/907333API
curl https://kongchang.com/api/v1/knowledge/claims/907333MCP
get_claim(id=907333)