Artificial intelligence foundational model company OpenAI is making Indic language performance and voice one of its top research priorities in India, according to Marc Manara, global head of startups at the IPO-bound tech firm.

“Last year we got so much feedback from founders on Indic language performance and voice specifically," Manara told Mint in an interview. “It's a big priority internally because we want to overcome this and it's important for our B2B customers and our consumer business.”

The focus on Indic voice comes as OpenAI works to expand its engagements with startups in India, which it counts among its most important. The firm has also cut the cost of some of its models to woo more customers in what it considers to be one of the more mature and vibrant startup ecosystems worldwide.

For both startups and venture capital investors, voice AI represents a massive opportunity in India, where spoken language remains the primary mode of communication. Industry experts bet that AI-driven voice will become the primary interface for how Indians interact with technology.

According to the IMARC Group, India’s conversational AI market was worth $653.24 million in 2025 and is expected to grow to $5.9 billion by 2034 at a compound annual growth rate of 25.6%.

OpenAI's approach to closing the gap on Indic languages and voice is rooted in a research-first strategy, one it has used before to make rapid gains in other categories such as image and code generation. “At the end of the day it's a research challenge and at OpenAI, if you point our researchers at a tough challenge, they're really good at producing amazing results,” said Manara. “I'm hopeful that we'll see more and more gains there but for this market it's one of the things that has come up very consistently.”

Cutting model costs

Last week, OpenAI slashed prices for its GPT-5.6 Luna and GPT-5.6 Terra models by 80% and 20%, respectively. The move comes as rising compute costs drive startups and enterprises toward open-weight alternatives, a shift that’s particularly pronounced in price-sensitive markets such as India.

According to Manara, the price cuts are intended to make AI viable for a wider range of applications that were previously too expensive for users. He acknowledged that while cheaper rates might reduce short-term revenue per client, making the models more affordable will ultimately drive broader adoption across more startups.

The price cuts also address a growing trend of startups turning to open-weight models to curb soaring compute bills. While Manara acknowledged this hybrid approach isn't new, he dismissed the idea that it poses a threat to OpenAI. He noted that most developers today use a multi-model approach, combining proprietary models from different AI providers with open-weight models depending on the specific task. He added that this combination is "healthy and kind of normal and good for the whole ecosystem”.

Rather than viewing churn as lost business, OpenAI said it treats startup defection to open-weight models as product feedback. If a developer opts for an open-weight alternative, the company sees it as a clear signal to fix gaps in its pricing or performance, rather than a threat to defend against.

Deepening ecosystem

OpenAI's startup team in India is the biggest across the Asia-Pacific and third largest globally. Just a year ago, the story was entirely different, because while the ecosystem knew the firm existed, nobody knew who to turn to for support.

Over the past year, OpenAI has tied up with several premiere venture capital funds and startups, running campaigns, hackathons, developer workshops and code sprints to make in-roads into India. That has largely worked, Manara said. “At the end of the day, we'd love for the next breakout startups in AI to be building and partnering closely with us.”

Over the past year, Indian founders have increasingly embraced Silicon Valley’s ‘move fast and break things’ ethos, according to Manara, who said this aggressive drive is the compounding result of earlier successes like Flipkart a decade ago. “It's like Silicon Valley 30 years ago, where you had the diaspora from Paypal who moved and created different companies, which then spawned the next set, like a compounding effect,” he said.

Artificial intelligence foundational model company OpenAI is making Indic language performance and voice one of its top research priorities in India, according to Marc Manara, global head of startups at the IPO-bound tech firm.

“Last year we got so much feedback from founders on Indic language performance and voice specifically," Manara told Mint in an interview. “It's a big priority internally because we want to overcome this and it's important for our B2B customers and our consumer business.”

The focus on Indic voice comes as OpenAI works to expand its engagements with startups in India, which it counts among its most important. The firm has also cut the cost of some of its models to woo more customers in what it considers to be one of the more mature and vibrant startup ecosystems worldwide.

For both startups and venture capital investors, voice AI represents a massive opportunity in India, where spoken language remains the primary mode of communication. Industry experts bet that AI-driven voice will become the primary interface for how Indians interact with technology.

According to the IMARC Group, India’s conversational AI market was worth $653.24 million in 2025 and is expected to grow to $5.9 billion by 2034 at a compound annual growth rate of 25.6%.

OpenAI's approach to closing the gap on Indic languages and voice is rooted in a research-first strategy, one it has used before to make rapid gains in other categories such as image and code generation. “At the end of the day it's a research challenge and at OpenAI, if you point our researchers at a tough challenge, they're really good at producing amazing results,” said Manara. “I'm hopeful that we'll see more and more gains there but for this market it's one of the things that has come up very consistently.”

Cutting model costs

Last week, OpenAI slashed prices for its GPT-5.6 Luna and GPT-5.6 Terra models by 80% and 20%, respectively. The move comes as rising compute costs drive startups and enterprises toward open-weight alternatives, a shift that’s particularly pronounced in price-sensitive markets such as India.

According to Manara, the price cuts are intended to make AI viable for a wider range of applications that were previously too expensive for users. He acknowledged that while cheaper rates might reduce short-term revenue per client, making the models more affordable will ultimately drive broader adoption across more startups.

The price cuts also address a growing trend of startups turning to open-weight models to curb soaring compute bills. While Manara acknowledged this hybrid approach isn't new, he dismissed the idea that it poses a threat to OpenAI. He noted that most developers today use a multi-model approach, combining proprietary models from different AI providers with open-weight models depending on the specific task. He added that this combination is "healthy and kind of normal and good for the whole ecosystem”.

Rather than viewing churn as lost business, OpenAI said it treats startup defection to open-weight models as product feedback. If a developer opts for an open-weight alternative, the company sees it as a clear signal to fix gaps in its pricing or performance, rather than a threat to defend against.

Deepening ecosystem

OpenAI's startup team in India is the biggest across the Asia-Pacific and third largest globally. Just a year ago, the story was entirely different, because while the ecosystem knew the firm existed, nobody knew who to turn to for support.

Over the past year, OpenAI has tied up with several premiere venture capital funds and startups, running campaigns, hackathons, developer workshops and code sprints to make in-roads into India. That has largely worked, Manara said. “At the end of the day, we'd love for the next breakout startups in AI to be building and partnering closely with us.”

Over the past year, Indian founders have increasingly embraced Silicon Valley’s ‘move fast and break things’ ethos, according to Manara, who said this aggressive drive is the compounding result of earlier successes like Flipkart a decade ago. “It's like Silicon Valley 30 years ago, where you had the diaspora from Paypal who moved and created different companies, which then spawned the next set, like a compounding effect,” he said.