Unverified50% confidenceFactExact time
语音Agent典型端到端延迟预算约为STT finalization 100-300ms + LLM TTFT 200-500ms + TTS TTFA 150-250ms + 网络开销50-100ms，总计500-1150ms
1
Sources
50%
Confidence
Long-term
Relevance
8/12/2026
First Seen
Sources
Related Claims
UnverifiedGPT-4-level models can have a Time to First Token (TTFT) of 500–2000ms, while smaller models such as Llama 3.1 8B can achieve a TTFT as low as 100–300ms.74% similarUnverifiedGPT-OSS 20B在RTX 3090上可达约40 Token/s的生成速度，在RTX 5090上能达到200 Token/s67% similarUnverified原生 PCIe 版本的 16G V100 TDP 约为 250W，而 SXM 版本的 TDP 为 300W65% similarUnverifiedRTX 4090的功耗约为300W65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/735048API
curl https://kongchang.com/api/v1/knowledge/claims/735048MCP
get_claim(id=735048)