Meta offers discounts for sharing AI model data
Meta has introduced a new pricing model for its Muse Spark AI models, offering significant discounts to users who agree to share their prompts and generated outputs for model development.
Meta is implementing a new pricing strategy for its Muse Spark AI models, providing substantial discounts to users who consent to share their prompts and model outputs for future development. This approach contrasts with the common practice of allowing users to opt out of data sharing for model training.
The new plan offers considerable cost reductions on token usage for customers who contribute their data. For instance, the input cost per million tokens, normally $1.25, is reduced to $0.10 for contributing users, an 8% rate. Similarly, the output cost per million tokens, typically $4.25, drops to $0.20 for those who share their data.
This move comes after Meta faced internal protests and eventually halted a plan earlier this year to track employee computer usage for data collection. The company has previously encountered challenges in acquiring sufficient training data for its AI models.
The data-sharing discount program aims to secure necessary training data for the ongoing refinement of Muse Spark models, including the recently released Muse Spark 1.3, which reportedly surpasses GPT-5.6 Sol in coding capabilities. The new pricing applies to Muse Spark 1.2 and 1.3 versions.