Grok 3 (xAI) and Llama Nemotron Ultra 253B (NVIDIA) are both frontier AI models. Grok 3 was published in February 2025 and Llama Nemotron Ultra 253B in March 2025.
Grok 3 was trained on 3.5 x 10^26 FLOP, about 8.9x the compute of Llama Nemotron Ultra 253B at 3.9 x 10^25 FLOP. Training compute is the closest available proxy for how much was invested in a model, though it says nothing on its own about how well that compute was spent.
These two models share no benchmark on which both have been scored, so no direct performance comparison is possible here. The specification table below is a comparison of inputs, not of results.
| Field | Grok 3 | Llama Nemotron Ultra 253B |
|---|---|---|
| Organization | xAI | NVIDIA |
| Published | Feb 17, 2025 | Mar 18, 2025 |
| Training compute | 3.5 x 10^26 FLOP | 3.9 x 10^25 FLOP |
| Parameters | 3T | 253B |
| Dataset size | — | 603B |
| Training hardware | NVIDIA H100 SXM5 80GB | — |
| Chips used | 80,000 | — |
| Training time | 2.2K h | — |
| Training cost (2023 USD) | $218M | — |
| Training power draw | 109.9 MW | — |
| Accessibility | API access | Open weights (restricted use) |
| Open weights | No | Yes |
| Country | United States | United States |
