OpenAI has announced that GPT-4, its latest text-generating model, is now generally available to existing API developers with a history of successful payments. Access will expand to new developers later this month, with availability limits increasing based on compute resources. GPT-4 offers improvements over its predecessor, GPT-3.5, including the ability to accept image inputs and perform at a human level on professional benchmarks. While not perfect, OpenAI plans to enable fine-tuning for GPT-4 and GPT-3.5 Turbo, and has also made DALL-E 2 and Whisper APIs generally available, while deprecating older models.
What is the current status of OpenAI’s GPT-4 model?
GPT-4 is now generally available through the OpenAI API.
Existing API developers with a history of successful payments can access it, with plans to expand access to new developers.
What are the key improvements of GPT-4 over GPT-3.5?
GPT-4 accepts image and text inputs, generates text and code, and performs at «human level» on benchmarks.
Unlike GPT-3.5, which only accepted text, GPT-4 has multimodal capabilities.
When will older OpenAI models be deprecated?
January 4, 2024.
GPT-3 and its derivatives will no longer be available and will be replaced with new base GPT-3 models.
Artículo Original: https://techcrunch.com/2023/07/06/openai-makes-gpt-4-generally-available/
Advices:
- If you’re an existing OpenAI API developer with a history of successful payments, start using GPT-4 through the API now.
- If you’re a new developer, prepare to apply for GPT-4 API access later this month.
- If you’re using older GPT-3 models, plan to upgrade your integrations to the new base GPT-3 models before January 4, 2024, or fine-tune replacements.
Deja una respuesta