OpenAI on Tuesday introduced new tools that make it easier for developers to access its most advanced voice feature, including what the company is calling a Realtime API.
The latest breakthrough technology, which has pleased developers but alarmed lawmakers and some ex-OpenAI execs, offers developers access to OpenAI’s most advanced voice feature, currently exclusive to ChatGPT. The new tools let any application incorporate human-like AI assistants that can make phone calls and use natural conversation to navigate complex software.
Previously, the process required three steps for developers — transcribing audio, running the generated-text model to come up with an answer to the query, and finally using a separate text-to-speech model.
“Real Time API,” known as Advanced Voice Mode in ChatGPT, lets developers easily incorporate voice with lower latency, and allows the user to interrupt the chat bot mid-sentence to guide the conversation.
At OpenAI’s second developer day on Tuesday, the company also lifted the curtain on a tool that would allow smaller models to learn from larger ones, along with “Prompt Caching” that cuts some development costs in half by reusing pieces of the text AI previously processed.
The accommodating changes for developers come as OpenAI shifts from a nonprofit to a traditional startup geared to raise funds, recruit talent and pump up revenue. OpenAI expects sales to more than double to $11.6 billion in 2025 from about $3.7 billion in 2024, according to a Reuters report. The company also just raised $6.6 billion that could value it at $157 billion.
But in its haste to grow, the company has endured withering criticism that its technology is moving too quickly and compromising the safety and security of consumers and companies.
Indeed, a Wall Street Journal report last week said breakneck growth is “tearing the company apart” and has led to a raft of upper-management departures in recent months, including those of chief technology officer Mira Murati and chief scientist Ilya Sutskever. On Tuesday, Durk Kingma, one of the OpenAI’s co-founders, said he was jumping to rival Anthropic.
At a press conference Monday, OpenAI said it was unphased by the departures.
“[Former chief research officer Bob McGrew] and Mira have been awesome leaders. I’ve learned a lot from them. They are a huge part of getting us to where we are today,” said OpenAI Chief Product Officer Kevin Weil, who joined in June. “And also, we’re not going to slow down.”
OpenAI has no real choice on that matter. The Microsoft Corp.-backed company is engaged in a sprint with Anthropic, Alphabet Inc.’s Google, Meta Platforms Inc., Apple Inc. and others to get its products into the hands of developers and establish market leadership.
