Unverified50% confidenceBenchmarkExact time
On the HardMatch benchmark, VGGT fails on 35% of the image pairs.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/27/2026
First Seen
Valid until: 1/25/2027
Sources
Related Entities
Related Claims
UnverifiedIn the IGG test, the strongest open-source vision-language model (VLM) achieved only 31.6% accuracy.60% similarUnverifiedOn the SimpleQA benchmark, gpt-5-thinking-mini declines to answer 52% of questions, with an error rate of 26%.60% similarVerifiedGGUF量化格式（如Q4_K_M）通过将权重精度从32位浮点压缩到4位整数，通常只带来约1-5%的性能损失59% similarUnverifiedWhen DINOv3 ViT-H+/16 is pruned to 50% sparsity, removing 427 million parameters, ImageNet-1k linear probe accuracy drops by only 0.8%, without weight correction or retraining.58% similarUnverifiedThe Qwen4 preview version scored only 35.9 on the HLE reasoning leaderboard, losing to a competitor's enhanced version that scored 40.57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1102691API
curl https://kongchang.com/api/v1/knowledge/claims/1102691MCP
get_claim(id=1102691)