Unverified60% confidenceFactExact time
Grok 4.5 是在数万块 NVIDIA GB300 GPU 上训练的
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
7/10/2026
First Seen
Valid until: 10/8/2026
Sources
Grok 4.5深度实测:编程能力媲美顶级模型,成本仅零头
youtubeTheo - t3․gg7/9/2026
Related Claims
VerifiedGrok 4.5在数万台英伟达GB300 GPU上完成训练,支持最高50万Token超长上下文,并原生支持视觉输入85% similarUnverifiedGrok 4.5基于全新的V9基础模型,参数达到1.5万亿,是上一代V8的三倍,使用数万张英伟达GB300 GPU训练82% similarVerified在配备 Tensor Core 的 NVIDIA GPU(如 A100、H100)上,FP16/BF16 矩阵运算的吞吐量可以达到 FP32 的 2-8 倍73% similarUnverified英伟达GB300 GPU是Blackwell架构旗舰型号,集成双颗B200 GPU芯片与Grace CPU,相比H100的FP8训练吞吐量提升约4倍71% similarUnverified英伟达GB300 Blackwell Ultra GPU集成两颗GPU芯片与一颗Grace CPU,通过NVLink-C2C互联,内存带宽可达8TB/s以上,推理性能相比H100提升4至5倍70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/427578API
curl https://kongchang.com/api/v1/knowledge/claims/427578MCP
get_claim(id=427578)