Topics
Latest
AI
Amazon
Image Credits:Justin Sullivan / Getty Images
Apps
Biotech & Health
Climate
Image Credits:Justin Sullivan / Getty Images
Cloud Computing
DoC
Crypto
endeavor
EVs
Fintech
fund-raise
Gadgets
Gaming
Government & Policy
Hardware
Layoffs
Media & Entertainment
Meta
Microsoft
privateness
Robotics
Security
Social
Space
startup
TikTok
exile
Venture
More from TechCrunch
Events
Startup Battlefield
StrictlyVC
newssheet
Podcasts
Videos
Partner Content
TechCrunch Brand Studio
Crunchboard
Contact Us
It ’s been a tumultuous week for OpenAI , full ofexecutive departuresandmajor fundraise development , but the inauguration is back at it , essay to win over developers to build shaft with its AI models at its 2024 DevDay . The company announced several unexampled tools Tuesday , including a public genus Beta of its “ Realtime API ” , for building apps with scurvy - latency , AI - father voice responses . It ’s not quite ChatGPT ’s Advanced Voice Mode , but it ’s close .
In a briefing with newsperson before of the event , OpenAI chief product officer Kevin Weil state the recent departures of chief technology officer Mira Murati and chief enquiry officer Bob McGrew would not affect the society ’s progress .
“ I ’ll start with saying Bob and Mira have been awe-inspiring leaders . I ’ve learned a lot from them , and they are a vast part of getting us to where we are today , ” tell Weil . “ And also , we ’re not going to slow down . ”
As OpenAI undergoes yet another atomic number 6 - suite redevelopment – a admonisher of theturmoil following last year ’s DevDay – the caller is endeavor to convince developer that it still volunteer the best chopine to build AI apps on . Leaders say the inauguration has more than 3 million developer build with its AI model , but OpenAI is control in an more and more competitive space .
OpenAI note it had cut cost for developer to access its API by 99 % in the last two years , though it was likely forced to by competitors such as Meta and Google incessantly undercutting their prices .
One of OpenAI ’s new features , dub the Realtime API , will give developer the luck to make almost real - time , spoken communication - to - speech experience in their apps , with the choice of using six voices provided by OpenAI . These voices are distinct from those offered for ChatGPT , and developer ca n’t practice third political party voices , so as to foreclose copyright issue . ( The articulation equivocally ground on Scarlett Johansson ’s is not uncommitted anywhere . )
During the briefing , OpenAI ’s head of developer experience , Romain Huet , share a demonstration of a trip planning app built with the Realtime API . The covering allowed user to verbally speak with an AI assistant about an upcoming trip to London , and get low - latency response . The Realtime API also has admission to a number of tools , so the app was able-bodied to gloss a map with eating place emplacement as it answered .
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
At another point , Huet showed how the Realtime API could verbalise on the headphone with a human to enquire about ordering food for an outcome . Unlike Google ’s notorious Duo , OpenAI ’s API ca n’t call restaurants or shops directly ; however , it can incorporate with call APIs like Twilio to do so . Notably , OpenAI isnotadding disclosures so that its AI model automatically identify themselves on yell like this , despite the fact that these AI - generated voices sounds quite realistic . For now , it seems to be the developer ’ responsibility to total this disclosure , something thatcould be required by a fresh California practice of law .
As part of its DevDay promulgation , OpenAI also introduced imagination fine - tuning in its API , which will let developer employ image , as well as text , to hunky-dory - melodic phrase their applications of GPT-4o . This should , in hypothesis , help developers improve the execution of GPT-4o for task involving visual savvy . OpenAI ’s head of product API , Olivier Godement , tells TechCrunch that developer will not be able to upload copyright imagination ( such as a picture of Donald Duck ) , images that depict violence , or other imagery that go against OpenAI ’s safety policies .
OpenAI is belt along to match what its competitors in the AI theoretical account licensing space already offer . Its prompt caching feature is standardised to the feature Anthropic launch several months agoallowing developer to cache frequently used context of use between API call , reducing cost and ameliorate latency . OpenAI say developers can save 50 % using this lineament , whereas Anthropicpromises a 90 % discountfor it .
Lastly , OpenAI is offering a fashion model distillation feature to countenance developers use larger AI models , such as o1 - prevue and GPT-4o , to fine - tune modest model such as GPT-4o miniskirt . Running smaller model generally provides cost preservation compared to running big ones , but this feature should rent developer improve the operation of those minuscule AI models . As part of model distillation , OpenAI is launch a beta rating dick so developers can measure their fine - tune ’s carrying into action within OpenAI ’s API .
DevDay may make openhanded waves for what it did n’t announce . For instance , there was n’t any news about the GPT Store announced during last year ’s DevDay . Last we ’ve heard , OpenAI has been piloting a revenue part political program with some of the most popular creators of GPTs , but the company has n’t shared much since then .
Also , OpenAI says it ’s not releasing any new AI model during DevDay this twelvemonth . Developers expect for OpenAI o1 ( not the preview or mini version ) or the startup ’s video generation mannequin , Sora , will have to look a little longer .
