Given the amount of compute required to calculate an answer (keep in mind that a 2000 token response is 2000 passes through the model) and the number of GPU’s needed to train the model, I’d guess they’re not making much. In fact it wouldn’t surprise me to learn that they’re at or below break even when they factor everything in.
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| OpenAI GPT-4 revenue estimate | 8 | 7735 | January 2, 2024 | |
| How much do tokens cost OpenAI? Or is it still unprofitable? | 4 | 1356 | January 2, 2024 | |
| GPT-4 Cost Estimate (UPDATED) | 5 | 13318 | January 2, 2024 | |
| Estimate the cost for 1 min usage of Real-time API | 5 | 3211 | January 9, 2025 | |
| How many api requests openai has to process per day? | 10 | 8466 | August 10, 2024 |