How much does Meta Muse Spark 1.3 cost?

The Standard tier costs $1.25 per million input tokens and $4.25 per million output tokens. The Contributor tier costs $0.10/$0.20 per million tokens but requires letting Meta train on your prompts and completions.

What is the Contributor tier?

It is a discounted pricing tier — up to 21x cheaper than Standard — where you allow Meta to use your prompts and completions for training in exchange for the lower rate.

Is Muse Spark 1.3 safe to use for sensitive data?

Only on the Standard tier, which Meta says does not train on your data. The Contributor tier explicitly involves Meta using your inputs and outputs for training, so it is not appropriate for confidential or regulated information.

What is Muse Spark 1.3's context window?

It supports a 1,048,576-token context window on both the Standard and Contributor pricing tiers.

How does Muse Spark 1.3 compare to GPT-6 Astra on price?

Muse Spark 1.3's Standard tier ($1.25/$4.25 per million tokens) is far cheaper than Astra's $10/$50 rate, though the two are not positioned as equivalent on raw capability — Astra targets frontier agentic and computer-use tasks, while Muse Spark 1.3 targets cost-efficient scale.

Meta Muse Spark 1.3: The Cheapest Frontier-Class Model Yet

Meta priced Muse Spark 1.3 at $1.25/$4.25 per million tokens, with a Contributor tier as low as $0.10/$0.20 in exchange for data training rights. Here is how the pricing actually works.

By Quantara Editorial Team · AI Tools · Published Sep 2, 2026 · 7 min read

While OpenAI and Anthropic were making headlines with frontier releases, Meta quietly shipped Muse Spark 1.3 on September 2, 2026, at a blended price near $0.10 per million tokens for its cheapest tier — undercutting nearly every other frontier-class model on the market. The release is less about raw capability and more about a pricing structure that trades cost for data rights, and it is worth understanding before assuming the headline number applies to you.

Muse Spark 1.3 Pricing: Two Tiers

Tier	Input / Output (per 1M tokens)	Trade-off
Standard	$1.25 / $4.25	Full data privacy — Meta does not train on your prompts or completions
Contributor	$0.10 / $0.20	Up to 21x cheaper, but Meta trains on your prompts and completions

Muse Spark 1.3 also has separate rates for cache reads, priced at $0.15 per million tokens, and web search at $2.50 per 1,000 calls. The model supports a 1,048,576-token context window at either pricing tier.

Why the Contributor Tier Exists

The Contributor tier is the mechanism behind Muse Spark 1.3's headline "cheapest frontier model" framing — that price only applies if you opt into Meta training on your data. For hobbyists, prototyping, and non-sensitive workloads, that trade may be an easy call. For any team handling customer data, proprietary code, or regulated information, the Standard tier's data-privacy guarantee is almost certainly worth the roughly 12x price premium over Contributor rates.

How It Compares to Recent Flagship Pricing

Muse Spark 1.3's Standard tier, at $1.25 input / $4.25 output, undercuts GPT-6 Astra's $10/$50 rate by a wide margin and comes in below Claude Fable 5.1's output pricing as well, though it is not positioned as a like-for-like competitor to either flagship on raw capability. Its Contributor tier goes further, landing well below even budget-oriented open-weight models on a pure price basis — the closest comparison is less "which model is smartest" and more "which model is viable to run at massive scale on high-volume, low-sensitivity tasks."

Who Should Use Muse Spark 1.3

High-volume, low-sensitivity tasks (bulk classification, simple summarization, internal tooling) are a strong fit for the Contributor tier's price.
Teams handling customer or proprietary data should default to the Standard tier despite the higher cost, or avoid Muse Spark's Contributor pricing entirely.
Cost-modeling teams should treat the $0.10/$0.20 headline figure as a best case, not the model's real price, when comparing it to competitors.

The Bottom Line

Muse Spark 1.3 is Meta's clearest statement yet that it intends to compete on price and scale rather than chase frontier benchmark leadership. The Contributor tier is a genuinely aggressive price point, but it is a data-for-cost trade, not a free lunch — teams should read the fine print before routing sensitive workloads through it, and should compare Meta's Standard tier, not its Contributor tier, when benchmarking against Astra or Fable 5.1 on a privacy-equivalent basis.

Sources

Frequently asked questions

How much does Meta Muse Spark 1.3 cost?: The Standard tier costs $1.25 per million input tokens and $4.25 per million output tokens. The Contributor tier costs $0.10/$0.20 per million tokens but requires letting Meta train on your prompts and completions.
What is the Contributor tier?: It is a discounted pricing tier — up to 21x cheaper than Standard — where you allow Meta to use your prompts and completions for training in exchange for the lower rate.
Is Muse Spark 1.3 safe to use for sensitive data?: Only on the Standard tier, which Meta says does not train on your data. The Contributor tier explicitly involves Meta using your inputs and outputs for training, so it is not appropriate for confidential or regulated information.
What is Muse Spark 1.3's context window?: It supports a 1,048,576-token context window on both the Standard and Contributor pricing tiers.
How does Muse Spark 1.3 compare to GPT-6 Astra on price?: Muse Spark 1.3's Standard tier ($1.25/$4.25 per million tokens) is far cheaper than Astra's $10/$50 rate, though the two are not positioned as equivalent on raw capability — Astra targets frontier agentic and computer-use tasks, while Muse Spark 1.3 targets cost-efficient scale.