Meta puts a price on data

Meta is offering an average 95% discount on its new Muse Spark model if users agree to share their prompts and outputs to train future AI models.

“Meta has taken that idea and put a price tag on it.”

Slashing token costs

Standard input tokens cost $1.25 per million and outputs cost $4.25. Under Meta's contributor tier, those prices drop to $0.10 and $0.20 per million tokens.

Meta's push for new training data

The aggressive discount follows Meta's recent struggles to source training data, including a controversial program tracking employee computer usage that was paused after internal pushback.

Agentic AI relies on workflow logs

Real-world interaction logs are essential for training AI agents. Major capability jumps in tools like coding assistants relied heavily on storing user sessions for reinforcement learning.

Pushing enterprises to share

Companies usually pay heavy premiums for enterprise plans to keep data private. Meta’s massive discount incentivizes firms to re-evaluate what data is genuinely proprietary.

“They stick with token-billed Enterprise plans even though the subscription-based consumer plans like Claude Max and ChatGPT Pro are discounted by 10x-20x”

Intensifying AI price wars

Meta's contributor tier ramps up price competition among frontier AI labs, following recent cost reductions from competitors like OpenAI and Anthropic.