Meta is offering a cheaper Muse Spark 1.2 “contributor” tier priced at $0.10/1M input and $0.20/1M output tokens in exchange for using user prompts for training

The company, pressed by investors to generate revenue from AI, says its offering will cost less than popular alternatives

Wall Street Journal 2026-08-05

Context & Ripple Effects

Meta had already positioned its model API around aggressive pricing and priced Muse Spark 1.1’s standard API tier at $1.25 per million input tokens and $4.25 per million output tokens. The contributor tier sharply separates price-sensitive workloads from customers that do not want their prompts used for training.

The timing also puts a low-cost data-sharing option beside Muse Code’s beta launch, which uses the coding-focused Muse Spark 1.2. Meta’s earlier coverage identified coding and agentic performance as priorities, making contributor prompts relevant input to the product line it is now commercializing.

First-order effects

Contributor-tier customers get input and output pricing far below Muse Spark 1.2’s standard rates, in return for allowing Meta to use their prompts for training.
Meta gains a new stream of training-eligible prompts while preserving a higher-priced standard tier for users that do not accept that trade-off.

Second-order effects

API buyers must segment workloads by data sensitivity: low-risk tasks can move to the contributor tier, while proprietary or regulated prompts remain on standard terms.
The tier makes Meta’s model API pricing strategy more explicitly a trade between cash cost and data rights, rather than a single price comparison with alternative models.

Third-order effects

If other model providers adopt comparable terms, prompt rights may become a priced input to model development, with privacy-sensitive inference carrying a premium.
The pattern shifts competition from headline token prices toward the effective cost of an AI task after accounting for whether the provider can retain and train on the underlying inputs.

The trend: AI model providers are increasingly using discounted inference to acquire training data while charging more for privacy-preserving usage.

Discussion

@kylebrussell Kyle Russell on x
like deepseek v4 flash 0731 this would basically feel free, sipping pennies
@giffmana Lucas Beyer on x
I think this is a very nice deal for random hobby stuff. In my experience, the model is an absolute beast in terms of execution. It's not quite as clever/clean as Sol/Fable in complex situations yet, but it easily holds up to Luna/Terra/Opus while being super fast. (At least in
@scaling01 @scaling01 on x
they actually give you the model almost for free if you allow them to train on your data [image]
@synthwavedd Leo on x
If you use their “contributor” slug - i.e. opt in to data collection for Meta to “improve their products” - you can use Spark 1.2 at ridiculous pricing of $0.10/Mtok input and $0.20/Mtok output [image]
@kimmonismus @kimmonismus on x
Really to see Meta catching up and continuing to release new titles quickly. Muse Spark 1.2 is a very good terminal agent that holds its own very well. The only caveat: it's being compared to Terra 5.6 xhigh instead of Sol high. [image]
@heaney555 David Heaney on x
Wow. Meta is providing a GPT Terra & Claude Sonnet class AI coding agent for 20x lower cost to developers if they'll allow the output to be used to train future models. That's a highly disruptive, though Faustian, offer.
@xeophon Florian Brand on x
@scaling01 Actually insane deal that might be very much worth it
@teortaxestex @teortaxestex on x
LMAO Muse Spark has cheaper input, output *and, uniquely, cache hits* than V4-Flash. They don't care about margins, they want the data

Chronicles