Alexandr Wang, chief artificial intelligence officer at Meta, said that the price of the newly released Muse Spark 1.3 will be the same as Muse Spark 1.2, but it is more efficient, and the tokens required to run are reduced by 25%. Meta has yet to decide whether to disclose the weights of Muse Spark 1.3, but the company still plans to disclose the weights of Muse Spark 1.2. The company is still developing a larger “watermelon” model as planned. “We believe that 'watermelon' will be extremely competitive.”

Alexandr Wang, chief artificial intelligence officer at Meta, said that the price of the newly released Muse Spark 1.3 will be the same as Muse Spark 1.2, but it is more efficient, and the tokens required to run are reduced by 25%. Meta has yet to decide whether to disclose the weights of Muse Spark 1.3, but the company still plans to disclose the weights of Muse Spark 1.2. The company is still developing a larger “watermelon” model as planned. “We believe that 'watermelon' will be extremely competitive.”