Unverified50% confidenceBenchmarkExact time
ElevenLabs的Turbo模型支持首句合成延迟约300ms
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/23/2026
First Seen
Valid until: 10/21/2026
Sources
Voice AI Agent Architecture: Pipecat Flows vs Vapi Squad — A Deep Comparison
redditr/aiagents7/12/2026
Related Claims
UnverifiedDeepgram Nova-3的流式识别延迟可低至100ms以内，ElevenLabs的Turbo模型支持首句合成延迟约300ms80% similarUnverified大模型调用是典型的I/O密集型操作，一次大模型API调用的延迟通常在1-30秒之间68% similarUnverifiedSmall model inference latency is approximately 500ms, while vector matching latency is only 30–100ms.67% similarUnverifiedWhen current mainstream large models handle complex code tasks, the time to first token (TTFT) is often between 2 and 8 seconds.66% similarUnverifiedThe latency for accessing overseas models via relay is typically 300–800 ms.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/599681API
curl https://kongchang.com/api/v1/knowledge/claims/599681MCP
get_claim(id=599681)