Currently it's $0.06/1k tokens, but OpenAI has a history of aggressively slashing API prices.
If OpenAI releases a different model in the GPT-4 family, it counts as long as it is qualitatively similar, by my subjective judgement. For example, gpt-3.5-turbo would count as similar to text-davinci-003. Criteria I'll use:
Is the context window at least 8K?
Is consensus that the model is a good replacement for current gpt-4 for most use cases?
If it is not possible to do inference on any GPT-4 family model through the OpenAI API on Jan 1, this resolves NO (because the price is effectively infinite).
I won't trade in this market.
1,000🏅 Top traders
| # | Trader | Total profit |
|---|---|---|
| 1 | Ṁ234 | |
| 2 | Ṁ127 | |
| 3 | Ṁ31 | |
| 4 | Ṁ29 | |
| 5 | Ṁ9 |
People are also trading
@DanMan314 this model would likely count (I'd want to verify quality), but it seems unlikely that the price for completion/output tokens will drop strictly below $0.02/1k by end of year