Topics
Latest
AI
Amazon
Image Credits:Didem Mente/Anadolu Agency / Getty Images
Apps
Biotech & Health
Climate
Cloud Computing
DoC
Crypto
Enterprise
EVs
Fintech
Fundraising
Gadgets
Gaming
Government & Policy
Hardware
Layoffs
Media & Entertainment
Meta
Microsoft
privateness
Robotics
surety
Social
Space
Startups
TikTok
Transportation
speculation
More from TechCrunch
event
Startup Battlefield
StrictlyVC
newssheet
Podcasts
Videos
Partner Content
TechCrunch Brand Studio
Crunchboard
adjoin Us
OpenAI is always making slight adjustments to its models and pricing , and today is no elision . The company hasreleased a handful of newfangled modelsand dropped the price of API access — this is primarily of interest to developer , but also serve as a bellwether for future consumer options .
GPT-3.5 Turbo is the model most people interact with , normally through ChatGPT , and it serve up as a kind of industry standard now — if your answers are n’t as near as ChatGPT ’s , why bother ? It ’s also a popular API , being lower price and quicker than GPT-4 on a lot of tasks . So pay users will be proud of to hear that input prices are dropping by 50 % and end product by 25 % , to $ 0.0005 per thousand relic in , and $ 0.0015 per thousand tokens out .
As people play with using these genus Apis for text - intensive coating , like analyzing entire papers or books , those tokens really start out to add up up . And as open germ or ego - managed manakin trip up up to OpenAI ’s performance , the fellowship needs to make indisputable its customer do n’t just leave . Hence the steady ratchet down of Leontyne Price — though it ’s also a natural resultant of streamlining the models and amend their infrastructure .
GPT-3.5 Turbo also gets a unexampled poser version , 0125 ( i.e. , today ’s date ) that include “ various improvement ” but apparently little that OpenAI thought deserving cite . The last version was 0613 , so it ’s a little surprising they do n’t mention more .
https://techcrunch.com/2024/1/23/chatgpt-everything-to-know-about-the-ai-chatbot/
GPT-4 Turbo gets a raw preview model for API use , also 0125 , which has an interesting fix :
This model completes tasks like computer code propagation more thoroughly than the previous preview framework and is mean to repress face of “ laziness ” where the mannikin does n’t fill in a task .
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
Perhaps the framework take the wrong deterrent example from the developers whose habit it ingested and embodied ! peradventure AI is already tranquil quitting . Good for AI !
This is still in preview manner . But GPT-4 Turbo with vision ( i.e. , GPT-4 V ) will set in motion to ecumenical availability “ in the coming month . ”
There are also a smattering of new and improved schoolbook embedding modelling , which are more for the people on the technical side . And the caller also issue a raw version of its innocent moderation API — which identifies potentially harmful text . Look for interpretation 007 if you ’re experiment with using this to augment your temperance want .
