New with GPT-4 Turbo

GPTs
Your customized ChatGPT for everything
Combines Instruction, Expanded Knowledge, Actions
Program it with your words
Share it publicly

GPT store: list your GPT, earn $ from people using your GPT
WOAH

New Model o’clock: GPT-4 Turbo
What’s new?

128k token context length 
(old was 8k or 32k)
That’s 300 pages of a book!!

JSON mode
Respond with valid JSON, giving more output control

Up to Date
Knowledge up to April 2023 (continuous improvement)
Bring in outside database knowledge

New Modalities
DALL-E 3,
GPT-4 Turbo w/ Vision
Text to speech
All going into API, today
Whisper V3 (open-source speech recognition) coming soon to API

Customization 
Extended to 16k version
Application for GPT-4 fine-tuning experimental access program

Custom Models:
researchers work with companies for cutting-edge info models

Higher rate limits
2x tokens per minute for all GPT-4 customers
Or request for even further limits

Copyright Shield
OpenAI pays if YOU face legal claims around copyright infringement 
(Both Enterprise And API)

GPT-4 Turbo Pricing
2.75x cheaper than GPT-4
3x less for input tokens than GPT-4 (1 cent per 1k)
2x less for completion than GPT-4 (3 cents per 1k)

No more ChatGPT Model picker menu
ChatGPT knows what model you need automatically

Takeaways:
OpenAI has done their homework on what the people WANT

AI will be:
More accessible
More democratized
Easier to use/build