Grok 3 vs Nemotron-4 340B

Grok 3 (xAI) and Nemotron-4 340B (NVIDIA) are both frontier AI models. Grok 3 was published in February 2025 and Nemotron-4 340B in June 2024.

Grok 3 was trained on 3.5×10²⁶ FLOP, about 19.4x the compute of Nemotron-4 340B at 1.8×10²⁵ 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.