How to choose an OpenAI Model
motive
OpenAI has many Model and frequent updates, so users don't know how to choose.
Earth models are systematically organized in this paper, so that Model selection can be quickly made Earth.
conclusion
If you want to see the conclusion directly, see the table below.
| classification | Training Data Cut-off Time | price | response speed | context length | other | Selected Model Name |
|---|---|---|---|---|---|---|
| GPT-4o | October month, annum | expensive | slow | 128K | gptgpt-4o | |
| GPT-4o-mini | October month, annum | inexpensive | fast | 128K | gptgpt-4o-mini | |
| Embedding | text-embedding-3-small | |||||
| 希望非英文的场景更精确 | text-embedding-3-large |
deadline
Data in this article is as of July 27, 2024 annum month day.
Model classification
| Model name | explain |
|---|---|
| GPT-4o | Wen Shengwen |
| GPT-4o-mini | |
| DALL·E | Vincent Graph |
| Sora | Wensheng Video |
| TTS | Wencheng Phonetic |
| Whisper | Phonetic recognition |
| Embeddings | |
| Moderation | censor content |
Embedding
| Model name | Description | vector Size | price |
|---|---|---|---|
| text-embedding-3-large | Suitable for non-English tasks. | 3,072 | $0.13 / 1M tokens |
| text-embedding-3-small | An upgraded version of text-embedding-ada-002 gaining a level. | 1,536 | $0.02 / 1M tokens |
| text-embedding-ada-002 | 1,536 | $0.10 / 1M tokens |