Grok 4.5 Training Scale Revealed
inductionheads · x · 2026-07-09
Reports indicate that Grok 4.5 is a 1.5T parameter model trained using only tens of thousands of GB300 units. The post also notes that Colossus 2 is already driving even larger-scale training capabilities, pointing to massive expansions in compute power and cluster size.
More from Infra
- Anthropic S-1 filing reveals 85% H1 gross margin and 5,300 employees — Jsevillamol · 2026-10-26
- HBM demand shifts: low-stack inference plans shelved, Rubin to go 12-high only — BenBajarin · 2026-10-26
- Anthropic generated $2.08 revenue per $1 of compute spend in Q2 2026, up from $0.63 — cedric_chee · 2026-10-26
- CircuitHub unveils The Grid, a factory-scale robot making PCBs in the US twice as fast as Shenzhen — garrytan · 2026-10-26
- Ubicloud publishes "Fixing Tool-Call Accuracy in vLLM" — petrusenko_max · 2026-10-26
- Meta approves nonvoting Class C shares as 2026 capex guidance hits $135B–$145B — thione · 2026-10-26