EXAONE 1.0 (LG) and Gemini 1.0 Ultra (Google DeepMind) are both frontier AI models. EXAONE 1.0 was published in December 2021 and Gemini 1.0 Ultra in December 2023.
Gemini 1.0 Ultra was trained on 5 x 10^25 FLOP, about 29.5x the compute of EXAONE 1.0 at 1.7 x 10^24 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 | EXAONE 1.0 | Gemini 1.0 Ultra |
|---|---|---|
| Organization | LG | Google DeepMind |
| Published | Dec 14, 2021 | Dec 6, 2023 |
| Training compute | 1.7 x 10^24 FLOP | 5 x 10^25 FLOP |
| Parameters | 300B | — |
| Training hardware | — | Google TPU v4 |
| Chips used | — | 57,000 |
| Training time | — | 2.4K h |
| Training cost (2023 USD) | $3M | $31M |
| Training power draw | — | 38.4 MW |
| Accessibility | Unreleased | API access |
| Open weights | No | No |
| Country | South Korea | United States |
