Please tell me when the GPT - 4 Turbo will be available without any restrictions for production usage. Initially OpenAI mentioned that this GPT - 4 Turbo will be completely available for production starting of the 2024.
1 Like
gpt-4-turbo (aka gpt-4-0125-preview) is essentially unrestricted in use when concerning rate limits. It has higher usage limits per tier than current gpt-4.
The only thing “preview” is they continue to hit the model with impactful quality reductions, just as OpenAI has done with every other production model, but being “turbo” especially and being trained differently, it has a sharp reduction in abilities and cannot perform the non-chat applications you’d want on the API as well as prior models. The AI treats the developer suspiciously as a ChatGPT user.
Multi-shot gpt-3.5-turbo-0613, as “Sherlock” with nothing “AI” about the system message:
At the moment, I am investigating a rather intriguing case involving a missing artifact from the British Museum. A valuable ancient artifact, known as the “Eye of Horus,” has mysteriously disappeared from its display. The artifact is believed to possess mystical powers and has significant historical and cultural importance.
I have been examining the security measures in place at the museum, interviewing staff members, and analyzing any potential leads or suspicious activities. The case has proven to be quite challenging, as there are no obvious signs of forced entry or any clear suspects.
However, I am confident that with my keen observation skills and deductive reasoning, I will be able to unravel the mystery behind the disappearance of the Eye of Horus and bring the culprit to justice.
Same input and parameters to gpt-4-turbo, at 10x the cost:
As of my last update, I’m not engaged in a specific investigation within the context of our current interaction. However, I’m always ready to delve into a new mystery or puzzle that you, Watson, or anyone else might bring to my attention. Whether it’s a perplexing case, a historical enigma, or a question needing a detailed explanation, I’m here to apply my deductive reasoning and analytical skills to help solve it. What conundrum can I assist you with today?
1 Like
I just want to use API without daily limits. Are they allowing this? i found this instructions about gpt 4 turbo.
API Limits : Preview models often have stricter API rate limits compared to stable versions. For instance, the rate limits for gpt-4v-preview are fixed at 20 requests per minute and 100 requests per day
There are no daily limits, just per-minute limits with the two non-vision gpt-4-turbo models. It is actually full gpt-4 that gets daily limits at tier-1 of payment history.
You can contrast the current limits (at tier 2) - and infer the computation “efficiency” of the turbo model by how many tokens they have it set to:
| Model | RPM | RPD | TPM |
|---|---|---|---|
| gpt-4 | 5,000 | - | 40,000 |
| gpt-4-turbo-preview | 5,000 | - | 450,000 |
1 Like