Unverified50% confidenceBenchmarkExact time
At a 50x compression ratio, KV-Surgeon improves question-answering accuracy by up to 4 percentage points over the attention-matching method (Zweiger et al., 2026) at comparable cost.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/20/2026
First Seen
Valid until: 1/18/2027
Sources
Related Entities
Related Claims
UnverifiedKV-Surgeon presents a faster variant that costs only 40% of the attention-matching method while still surpassing existing state-of-the-art methods.75% similarUnverifiedgpt-oss-120b spent about 25 times the cost per correct answer on the distractor counting task, while its accuracy was 7.5 percentage points lower (not significant).65% similarUnverifiedOutput token pricing is typically 2–4 times that of input tokens, because the generation process requires step-by-step decoding and computation of attention weights.64% similarUnverifiedWhen using the multi-agent verification process, MARCH's accuracy was suppressed to 4.4%, whereas when the same model was directly asked which solution was better, accuracy reached 43.7%.64% similarUnverifiedIn the advisor pairing mode, each 1 percentage point improvement in performance requires paying about 1.24 times more cost.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1073863API
curl https://kongchang.com/api/v1/knowledge/claims/1073863MCP
get_claim(id=1073863)