72,816 t
#1 of 13
estimated CO₂-equivalent emissions from training Grok 4 (2025), more than 1,000 average cars' lifetime emissions
What the chart shows
Training AlexNet in 2012 produced an estimated 0.01 tonnes of CO₂ equivalent. Training Grok 4 in 2025 produced about 72,816 tonnes. The average car emits about 63 tonnes over its lifetime, so Grok 4’s training run equals the lifetime emissions of more than 1,000 cars.
Emissions depend on more than model size: hardware efficiency, training time and the carbon intensity of the local electricity all matter (see the related chart).
Key numbers
| Measure | Value |
|---|---|
| Grok 4 (2025) | 72,816 t |
| Grok 3 (2025) | 59,200 t |
| Llama 3.1 405B (2024) | 8,930 t |
| GPT-4 (2023) | 5,184 t |
| AlexNet (2012) | 0.01 t |
How to read it
- Estimates by the AI Index, not reported by developers.
- The horizontal axis is logarithmic, so each gridline is ten times the one before.
- Car comparison: average lifetime emissions of 63 tonnes (Strubell et al., 2019).
Source: AI Index, 2026; Strubell et al., 2019, via the AI Index Report 2026 (Stanford HAI), figure 1.4.3: Estimated carbon emissions from training select AI models and real-life activities, 2012–25. Chart and text by AI Adoption.
Numbers checked against the published data on 19 October 2026.
Download the data (CSV) · Cite this chart
Numbers checked against the published data on 19 October 2026.
Download the data (CSV) · Cite this chart
Discussion