Expired60% confidenceFactExact time
Grok 4.5 was trained on tens of thousands of NVIDIA GB300 GPUs.
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
7/10/2026
First Seen
Valid until: 10/8/2026(expired)
Sources
Grok 4.5 Deep Dive: Coding on Par with Top Models, at a Fraction of the Cost
youtubeTheo - t3․gg7/9/2026
Related Claims
VerifiedGrok 4.5 was trained on tens of thousands of NVIDIA GB300 GPUs, supports an ultra-long context of up to 500,000 tokens, and natively supports visual input.85% similarUnverifiedGrok 4.5 is based on the all-new V9 base model, with parameters reaching 1.5 trillion—three times those of the previous-generation V8—and was trained using tens of thousands of NVIDIA GB300 GPUs.82% similarUnverifiedThe Grok model series has iterated from Grok-1 to Grok-2 and then to Grok-3, with Grok-3 claiming to have been trained using more than 100,000 H100 GPUs when it was released in early 2025.72% similarUnverifiedThe NVIDIA GB300 GPU is the flagship model of the Blackwell architecture, integrating dual B200 GPU chips with a Grace CPU, delivering approximately 4x the FP8 training throughput compared to the H100.71% similarVerifiedNVIDIA's GB300 Blackwell Ultra GPU integrates two GPU chips and one Grace CPU, interconnected via NVLink-C2C, with memory bandwidth exceeding 8 TB/s and inference performance 4 to 5 times higher than the H100.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/427578API
curl https://kongchang.com/api/v1/knowledge/claims/427578MCP
get_claim(id=427578)