can someone help with replacement model name in GPT-4 model series for gpt-3.5-turbo-16k similar parameters like max tokens and etc…
1 Like
Hi there - why GPT-4 replacement?
The nearest alternative for gpt-3.5-turbo-16k would be gpt-3.5-turbo-1106 or gpt-3.5-turbo-0125, both coming with a 16k context window.
Perhaps I misunderstand the question?
The closest GPT-4 model in terms of capabilities is gpt-4-0613, which also has function call ability, and has an 8k flexible context window length. There is also a gpt-4-0613 with 32k context length that an elite few have access to, which has high capabilities at the highest cost.
The “budget AI” GPT-4 would be gpt-4-1106-preview, which is different in that it has been trained on and makes available parallel tool use, and is a “turbo” model of distinctly lower quality and 1/3 the cost. The context length is longer for input, but it will refuse to write long output by training and the output token limit of 4k will never be reached.