Meta has introduced a contributor pricing tier for its Muse Spark model that sharply reduces costs for users who agree to share prompts and model outputs for future training. Muse Spark is described as a model intended for operating coding and other agents.
Contributor pricing and token rates
Under the standard agreement, 1 million input tokens cost $1.25; on the contributor tier those same input tokens cost $0.10. For output tokens the standard price is $4.25 per million, while contributor pricing lists $0.20 per million. The discounts average out to about 95% for users who “contribute” usage data.
Meta did not respond to a request for comment about the new pricing model. The company has previously faced internal pushback over data collection: an initiative to track employee computer usage was paused in June after attracting criticism.
Why usage data matters
Developer and open source maintainer Mario Zechner, who works on the Pi harness, attributed rapid capability gains in coding agents to stored session data. He said the jump in coding agent performance between April 2025 and October 2025 was driven in part by Claude Code’s default of storing coding agent sessions and using them for reinforcement learning training.
Princeton computer scientist Arvind Narayanan wrote on social media that large companies generally avoid having their data used for model training and instead choose token-billed Enterprise plans, even when consumer subscription plans such as Claude Max and ChatGPT Pro are discounted by 10x–20x. Narayanan suggested Meta’s paid contributor approach could encourage firms to separate proprietary data from material they are willing to share.
Meta’s pricing guide frames the contributor tier as a way to “lower the barrier to entry for prototyping, testing integrations, and scaling experiments where training on your data is acceptable.” The move arrives as competitors have also adjusted prices: Anthropic lowered costs for processing cached tokens alongside the release of Fable and Mythos, and OpenAI implemented price cuts at the end of July.
Original source: TechCrunch AI