Unverified50% confidenceBenchmarkExact time
On the test machine (Intel i5-13420H, 16 GB RAM, no discrete GPU), Laya-Heart's card median latency was 0.44 seconds, the first full reply 3.6 seconds, and two replies 5.8 seconds; on an idle machine these were 0.39 seconds, 3.5 seconds and 5.0 seconds respectively.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/18/2026
First Seen
Valid until: 1/16/2027
Sources
Related Entities
Related Claims
VerifiedOn a T4 GPU, Laya takes 33 milliseconds per single question, and batch processing can compress that to 7 milliseconds79% similarUnverifiedTested on the same machine equipped with a Spark CPU and limited to three threads, Laya's median call latency was 541 milliseconds and Kev's was 756 milliseconds, with the 95th percentile at approximately 607 and 802 milliseconds respectively.77% similarUnverifiedMatrix-Game 4.0 has an end-to-end latency of 112 milliseconds, whereas Code World Model is about 10.2 seconds, StreamRender-H3 about 545 milliseconds, and AgentGarten about 878 milliseconds.74% similarUnverifiedOn the same device, simply removing the 30fps frame rate cap during the preloading phase reduced load time from 4.1 seconds to 1.7 seconds.73% similarUnverifiedBEINGS的冷启动加载时间分别为7.5秒（SmolLM2 273MB）、13.4秒（Qwen 483MB）、26.8秒（Llama 1238MB），缓存后降至1.0、2.2、3.1秒。72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1062478API
curl https://kongchang.com/api/v1/knowledge/claims/1062478MCP
get_claim(id=1062478)