Expired60% confidenceBenchmarkExact time
On a 2000-token new prompt, DeepSeek V4.1 Flash at Q4 quantization starts prompt processing at approximately 360 tokens/s, dropping to approximately 307 at 64K context; token generation is stable at approximately 14 tokens/s.
2
Sources
60%
Confidence
Short-term (~14 days)
Relevance
10/5/2026
First Seen
Valid until: 10/19/2026(expired)
Sources
Related Entities
Related Claims
VerifiedDeepSeek-V4.1-Flash's KV cache is compressed to 890 bytes per token, only one-quarter that of V4 Flash.77% similarUnverifiedGLM 5.3 Flash is overall slower, with Q2 quantization generating approximately 12 tokens/s (approximately 10 at 64K).75% similarUnverifiedDeepSeek V4 Flash is a cloud-based model focused on high throughput, with an inference speed of approximately 140 tokens per second.75% similarUnverifiedDeveloper Primitive reported a generation speed of 142.6 tokens/s for Qwen3 Flash Next running locally.74% similarUnverifiedOfficial benchmark results show that the Qwen3.6 27B dense model with NVFP4 quantization achieves 202 token/s at single concurrency, 400 token/s at dual concurrency, and 1147 token/s at 8 concurrency.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/977949API
curl https://kongchang.com/api/v1/knowledge/claims/977949MCP
get_claim(id=977949)