Meta Platforms, Inc. announced on Sept. 3, 2026 a new “contributor” pricing tier for its Muse Spark model that reduces the cost of input tokens from $1.25 to $0.10 per million and output tokens from $4.25 to $0.20 per million. Meta says the tier delivers an explicit discount averaging about 95% for users who share their prompts and model outputs with the company for future development.
Pricing details and discount calculation
The standard pricing for Muse Spark, as reported by TechCrunch, is $1.25 per million input tokens and $4.25 per million output tokens. Under the contributor tier the same volumes cost $0.10 and $0.20 respectively. The per‑token discount percentages are shown in the table below.
| Token type | Standard price (USD per million) | Contributor price (USD per million) | Discount % |
|---|---|---|---|
| Input | 1.25 | 0.10 | 92% |
| Output | 4.25 | 0.20 | 95% |
Meta describes the overall discount as “about 95%” for data‑sharing users. The calculation is straightforward: the combined standard cost for one million input tokens and one million output tokens is $5.50, while the contributor tier charges $0.30, a reduction of roughly 95%.
Why the tier matters for developers
The pricing change is positioned as an incentive for developers who are willing to let Meta ingest the prompts they send to Muse Spark and the model’s generated outputs. By aggregating that data, Meta can train future iterations of its agentic tools. The discount therefore ties lower cost to a data‑sharing agreement, a model that other AI firms have explored but have not priced as aggressively.
For developers building coding assistants, autonomous agents, or other applications that consume large token volumes, the contributor tier could cut operating expenses dramatically. A developer running 10 million input tokens and 20 million output tokens per month would pay $52.50 under standard rates but only $2.70 under the contributor tier – a saving of $49.80 each month.
Meta has not disclosed how many customers have opted into the tier, nor the duration of the discount. The company’s filing history shows no prior mention of a contributor pricing model, making this the first public data point on Meta’s token‑price strategy for Muse Spark.
Meta’s scale and financial backdrop
Meta remains one of the largest U.S. technology companies. Its most recent Form 10‑Q, filed July 30, 2026, reports $38.9 billion in revenue for fiscal year 2018 (the only historic revenue figure in the packet) and a net income of $42.6 billion for the six months ended June 30, 2026. Total assets stand at $449.956 billion and shareholders’ equity at $261.221 billion as of the same date. While these figures pre‑date the Muse Spark pricing announcement, they illustrate the financial depth from which Meta can experiment with pricing models.
The company is listed on the Nasdaq under ticker META and files with the SEC under CIK 1326801. Its SIC description is “Services‑Computer Programming, Data Processing, Etc.” The packet does not provide a current chief executive, headquarters location, employee headcount, or founding year; those details should be verified against the latest company filings before publication.
Timeline and next steps
- Sept. 3, 2026 – TechCrunch publishes an article detailing the new contributor pricing tier for Muse Spark.
- Sept. 3, 2026 – Meta’s public announcement (as captured by TechCrunch) includes the exact token‑price figures and the stated overall discount of about 95%.
The announcement arrives amid a broader industry trend of AI providers lowering token prices to attract developers and gather usage data. While the packet does not contain statements from competitors, the pricing contrast with Anthropic and OpenAI’s recent cuts suggests a competitive environment where data access is becoming a differentiator.
What remains unknown is how Meta will enforce the data‑sharing requirement, whether the discount is permanent or limited to a trial period, and how the tier will affect Meta’s overall revenue from AI services. The company has not released a separate press release, so the TechCrunch article is the sole source for the pricing details.
Analysts will likely watch adoption rates closely. If a significant share of Muse Spark users opt into the contributor tier, Meta could amass a sizable corpus of real‑world usage data, potentially accelerating the development of next‑generation agentic tools.
For readers tracking AI‑model economics, the key takeaway is that Meta is now pricing token usage for a specific model in direct proportion to the data it receives, offering a discount that, on paper, reduces costs by up to 95% for participating developers.