Verified65% confidenceFactExact time

Meta used more than 16,000 H100 GPUs to train Llama 3.1 405B, with training taking several months and costs potentially exceeding $100 million.

3

Sources

65%

Confidence

Long-term

Relevance

9/4/2026

First Seen

Sources

Quasar 438B: Europe's New AI Sovereignty Flagship — A Deep Dive into the 438-Billion-Parameter Model

Related Entities

Metacompany Llama 3.1 405Bmodel H100product

Related Claims

VerifiedIndustry estimates put the training cost of GPT-4 at over $100 million, involving tens of thousands of A100 GPUs, months of continuous computation, and petabyte-scale training data.78% similar Unverified训练DeepSeek-V3仅消耗约278万H800 GPU小时，成本约合550万美元73% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/852857

API

curl https://kongchang.com/api/v1/knowledge/claims/852857

MCP

get_claim(id=852857)