Unverified50% confidenceBenchmarkExact time
The research team spent $42.57 on large-scale experiments across 23,440 scoring episodes, using a pre-registered 3-model × 3-random-seed grid design involving GPT-4.1 and Claude Haiku 4.5.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/11/2026
First Seen
Valid until: 12/10/2026
Sources
Related Entities
Related Claims
Unverified2.4万亿参数模型在BF16精度下考虑KV Cache和激活值后实际约需1.4-1.5TB显存60% similarUnverified研究表明当模型参数规模达到3B以上时，1.58-bit模型在多项基准测试中表现可接近甚至匹配同等规模的全精度模型60% similarVerifiedTongyi Qianwen released the Qwen 3.8 Max model, with a parameter scale as high as 2.4 trillion.60% similarUnverified千问3是阿里云通义实验室推出的开放权重模型系列，覆盖0.6B至235B参数规模，支持混合推理模式且许可证允许商业使用60% similarUnverifiedA China Academy of Information and Communications Technology (CAICT) special MCP testing report shows that a 27B local large model released by Shanghai Yuandian Xinhui ranked second in a comprehensive multi-tool collaboration evaluation, with a total score surpassing lightweight models with more than 200 billion parameters.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/892956API
curl https://kongchang.com/api/v1/knowledge/claims/892956MCP
get_claim(id=892956)