Meta Offers Discounts for Users Who Share AI Model Usage Data

Meta is introducing a new pricing model for its Muse Spark AI model that offers significant discounts to users who allow the company to use their prompts and model outputs to improve future AI systems.

The contributor pricing option is designed for users who are willing to share usage data with Meta during model development. Under this model, input and output token costs are reduced compared with standard pricing, lowering the cost for companies and developers testing AI applications.

According to Meta’s pricing documentation, one million input tokens cost $0.10 under the contributor model compared with $1.25 under standard pricing. Output tokens are priced at $0.20 per million under the contributor option, compared with $4.25 under standard pricing.

Meta seeks more data to develop AI agents

AI companies increasingly rely on real-world usage data to improve agent-based systems, but obtaining high-quality workflow data remains challenging. Meta previously faced internal criticism over an initiative designed to track employee computer usage, which was later paused. Reuters reported on Meta’s decision to pause the internal tracking effort.

Developers and researchers have argued that access to real user interactions can help improve AI agents, particularly for tasks involving coding and complex workflows. However, many organisations are cautious about allowing sensitive business data to be used for model training.

Princeton computer science professor Arvind Narayanan noted that many companies continue to choose enterprise AI plans with stronger data controls despite higher costs, reflecting concerns about data retention and governance. He discussed these issues in a social media post.

AI companies compete on pricing and data access

Meta’s contributor pricing approach creates a direct trade-off between lower AI costs and model providers’ ability to access usage data for development. The company says the option is intended for prototyping, testing integrations, and scaling experiments where users are comfortable sharing information.

The move comes as major AI companies continue adjusting pricing strategies. OpenAI has also introduced lower-cost models, while other AI providers have reduced prices for certain workloads as competition in the market increases.