Unverified50% confidenceBenchmarkExact time
In benchmarks on Amazon EC2 P5en instances with AWS EFA, the push-based KV transfer scheme achieved about a 30% improvement in TPOT (time per output token).
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/19/2026
First Seen
Valid until: 11/2/2026
Sources
Related Entities
Related Claims
UnverifiedOn Amazon EC2 P5en instances with EFA networking, Qwen3-32B achieved up to a 1.6x improvement in TTFT.67% similarUnverifiedThrough request-owned speculative KV, in a DeepSeek-V4-Pro 1P1D, MI355X, concurrency-256 scenario, the average TTFT was reduced from 19.4 seconds to 15.1 seconds.60% similarUnverifiedHaiku 5.5 can reach 100 to 200 TPS across different providers, and the creator's tests using Claude Code consistently showed approximately 180 TPS.58% similarUnverifiedvLLM's bidirectional KV transfer reduces TTFT in a 24k token scenario on Qwen3-32B from 778ms to 460ms, about a 1.6x improvement.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1070048API
curl https://kongchang.com/api/v1/knowledge/claims/1070048MCP
get_claim(id=1070048)