Meta Muse Pricing Drops 29% Below DeepSeek via Data Sharing Tier

Meta introduces Contributor tier for Muse Spark 1.2 and Code, cutting costs by 29% against DeepSeek V4 Flash in exchange for training data rights.

Woofun AI reports that Meta has established a Contributor discount tier for Muse Spark 1.2 and Muse Code, allowing users to grant permission for prompt and output usage in model training in exchange for reduced fees. Under this structure, input, cache hit, and output costs are set at $0.10, $0.002, and $0.20 per million tokens, respectively, representing a 29% reduction compared to DeepSeek V4 Flash rates of $0.14, $0.0028, and $0.28.

Users opting out of data sharing must utilize the standard tier, where prices rise to $1.25, $0.15, and $4.25, marking increases of 12.5x, 75x, and 21.25x over the Contributor tier. The Contributor tier also imposes stricter limits, capping requests at 60 per minute and 2.1 million tokens, whereas the standard tier allows 3,000 requests per minute and 4 million tokens. DeepSeek has indicated upcoming API price hikes, while Meta’s aggressive pricing positions it as a cost leader.