Amazon Titan (Amazon) and GPT-3.5 (davinci-002) (OpenAI) are both frontier AI models. Amazon Titan was published in September 2023 and GPT-3.5 (davinci-002) in March 2022.
Amazon Titan was trained on 4.8 x 10^24 FLOP, about 1.9x the compute of GPT-3.5 (davinci-002) at 2.6 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 | Amazon Titan | GPT-3.5 (davinci-002) |
|---|---|---|
| Organization | Amazon | OpenAI |
| Published | Sep 28, 2023 | Mar 15, 2022 |
| Training compute | 4.8 x 10^24 FLOP | 2.6 x 10^24 FLOP |
| Parameters | 200B | — |
| Dataset size | 4T | — |
| Training hardware | NVIDIA A100 | NVIDIA A100 SXM4 40 GB |
| Chips used | 13,760 | — |
| Training time | 1.2K h | — |
| Training cost (2023 USD) | $8M | $5M |
| Training power draw | 10.9 MW | — |
| Accessibility | API access | API access |
| Open weights | No | No |
| Country | United States | United States |
