Verified75% confidenceFactExact time
Inference latency is typically decomposed into two core metrics: time to first token (TTFT) and subsequent token generation throughput.
3
Sources
75%
Confidence
Long-term
Relevance
7/13/2026
First Seen
Sources
The Rise of Diffusion Language Models: DiffusionGemma Crushes Autoregressive Models in Speed Tests
redditr/GoogleGeminiAI7/11/2026
Related Claims
Unverified对流式响应场景，应重点关注首Token延迟（TTFT）而非整体完成时间81% similarUnverified传统混合部署中Prefill阶段的大批量计算会阻塞Decode阶段的token生成，导致TTFT波动剧烈76% similarUnverifiedTwo common metrics for evaluating local model speed are time to first token (TTFT) and generation speed (tokens/s).72% similarUnverifiedLatency and throughput are two different technical metrics.71% similarUnverifiedtokens/s指标可细分为首token延迟（TTFT）和生成速率（Generation Throughput）两个子维度71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/497411API
curl https://kongchang.com/api/v1/knowledge/claims/497411MCP
get_claim(id=497411)