Murf AI launches Falcon 2, a lower-cost voice AI model that takes on OpenAI and ElevenLabs; model starts audio in under 100ms at $0.01 per minute

Murf AI has launched Falcon 2, a new voice AI model that aims to compete with industry leaders like OpenAI and ElevenLabs. The model is designed to generate audio in under 100 milliseconds and is priced at $0.01 per generated minute, significantly lower than ElevenLabs' $0.05 rate.¹³

According to co-founder and CEO Ankur Edkie, the primary focus of Falcon 2 was to enhance voice quality while maintaining the low latency and cost profile of its predecessor. Edkie stated, “We are still the most cost effective and the lowest latency model out there globally,” emphasizing the model's competitive edge.

The model ranks ahead of OpenAI's Realtime and ElevenLabs' Flash and Turbo models for naturalness, as per the independent Artificial Analysis Speech Arena. It supports over 150 voices across more than 35 languages, including 12 Indian languages.²

Murf has opted for a lightweight proprietary architecture to improve inference efficiency without increasing model size. This approach allows the company to use smaller, commodity GPUs, which helps manage infrastructure utilization as customer demand fluctuates. “The main innovation from Falcon 1 to Falcon 2 was improving quality without increasing the amount of compute required,” Edkie noted.

At the quoted price of $0.01 per generated minute, generating one million minutes of speech would cost $10,000, compared to $50,000 for ElevenLabs' models. Murf is also exploring on-premise deployments for BFSI customers in India and the US, further expanding its market reach.

Murf has raised $11.5 million from investors and is considering a Series B funding round later this year, although an IPO is not an immediate priority as the company focuses on expanding its product offerings.⁵