Unverified50% confidenceBenchmarkExact time
Directly querying the VLM yielded a four-label exact-match accuracy of only 22.6%.
1
Sources
50%
Confidence
Long-term
Relevance
10/5/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe text-only memory method achieved a four-label exact-match accuracy of only 16%–17%.70% similarUnverifiedThe locally run Gemma 26B achieved an accuracy of only 49.2% across 453 scored predictions, while its paired always-predict-up baseline had an accuracy of 53.4%.69% similarUnverifiedState-of-the-art supervised models on the LIAR benchmark still hover around 28%–32% six-class accuracy.67% similarUnverifiedOn the legacy codebase test leaderboard, most models score only in the teens to twenties; Sol ranks fourth based on one recorded entry (which is not the same result as this test).67% similarUnverified在GPT-4发布初期，主流模型在GPQA上的准确率约为35-40%，人类领域专家基准线约为65%，随机猜测为25%66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/975940API
curl https://kongchang.com/api/v1/knowledge/claims/975940MCP
get_claim(id=975940)