Last verified
LATEST CHECKPOINT1M CONTEXTAGENTIC + CODINGNO LONG-CONTEXT PREMIUMAUDIO DEGRADED

Muse Spark 1.3 API Pricing

Muse Spark 1.3 is the newest checkpoint on Meta's Model API, described by Meta as tuned for agentic workflows — multi-step tool, browser and long-horizon tasks — with improved coding over 1.2, and marked recommended for new work. It is priced identically to 1.2 and 1.1: Meta groups all three and states they share the same standard pricing$1.25/M input, $0.15/M cached input, $4.25/M output. One thing to read before switching: Meta says audio understanding here is not fully supported and points audio work back to 1.2.

Input - per 1M tokens
$1.25/M
Source ai.developer.meta.com launch rate
Output - per 1M tokens
$4.25/M
3.4x the input rate launch rate
Cached input - per 1M tokens
$0.15/M
No cache-write charge published -88%
Effective - agentic blend
$0.66/M
92/8 split - 82% cache
§ 01 / TERMINAL

Run the numbers.

Live calculator on Meta's standard rate. The lever that actually moves this bill is not the workload mix but the tier: the Contributor rate below is a different order of magnitude, and it is paid for with training rights rather than dollars.

$ /mo
Workload split
Prompt cache hit rate
Tokens you can process
Words equivalent (English)
Effective rate
Open full calculator (all models · share URL · CSV) →
§ 02 / SCENARIOS

Real-world presets.

§ 03 / TOKENIZER

Paste text. See tokens. See cost.

Estimate · meta-tokenizer-estimate · ≈3.85 chars/token Auto-counts as you type

This is a chars-per-token approximation, not a real tokenizer. Actual tokens vary by language, code density, and tool-call overhead — counts are typically ±10–20% off for English prose, more for code or non-Latin scripts. For exact billing, use the vendor's official tokenizer.

Characters 920
Words 148
Tokens (estimated) 239 tokens
Cost as input · uncached $0.0003 USD
Cost as output · uncached $0.00102 USD
Cost as cached input $0.00004 USD
§ 04 / SHELF

Up against the shelf.

All models →
Model Input /M Output /M Effective blended Context Best for
Muse Spark 1.3 Current $1.25 cache $0.15 $4.25 $0.66 agentic 92/8 1M Agentic workflows and coding - recommended for new work
Muse Spark 1.2 $1.25 cache $0.15 $4.25 $0.66 identical card, previous checkpoint 1M Audio work, where 1.3 is degraded
Muse Spark 1.1 $1.25 cache $0.15 $4.25 $0.66 identical card, original checkpoint 1M No Contributor tier on this one
Gemini 3.5 Flash-Lite $0.30 cache $0.03 $2.50 $0.272 budget 1M peer 1M High-volume 1M-context work on a budget
Grok 4.3 $1.25 cache $0.20 $2.50 $0.558 same input, cheaper output 1M Output-heavy 1M-context agents
GPT-5.6 Luna $0.20 cache $0.02 $1.20 $0.152 OpenAI 1M peer 1.05M OpenAI-native agent stacks
Gemini 3.6 Flash $0.75 cache $0.075 $3.75 $0.481 Google agentic peer 1M Multimodal agents with computer use
Kimi K3 $3.00 cache $0.30 $15.00 $1.92 frontier agentic peer 1M Long-horizon agentic coding

Frequently asked.

What the Contributor tier really costs you, why three checkpoints share one price, and the audio footnote worth reading before you switch.

Q · 01 How much does Muse Spark 1.3 cost? +
Meta lists $1.25/M input, $0.15/M cached input and $4.25/M output on the Standard tier. Under this site's 92/8 agentic blend at an 82% cache-hit rate that is $0.66/M effective. There is no long-context premium — Meta states the rate is the same whether the window is nearly empty or nearly full.
Q · 02 Is it more expensive than 1.2 or 1.1? +
No. Meta groups all three checkpoints under one heading and says they "share the same standard pricing". Three versions, one card — so moving to 1.3 costs nothing and the decision is about capability. Meta positions 1.3 for agentic workflows and better coding, and calls it recommended for new work.
Q · 03 What is the Contributor tier, and what does it actually cost? +
It is the same checkpoint at $0.10/M input, $0.002/M cached input and $0.20/M output — about 19x cheaper on our blend, $0.034/M against $0.66/M. The price is not money: Meta describes it as "heavily discounted token pricing in exchange for permission to use your prompts and completions to train future Meta models". Throughput differs too — Contributor is capped at 100 requests per minute against Standard's 3,000, though token throughput is close (3M vs 4M per minute). It suits prototyping and load testing; anything touching customer or proprietary data is a legal question, not a pricing one. Only 1.3 and 1.2 offer it — 1.1 does not.
Q · 04 Can I send it audio? +
Technically yes, but Meta advises against it. Its modality table footnotes that "audio understanding in Muse Spark 1.3 is currently not fully supported, and response quality for requests including audio content may be degraded", and points you to Muse Spark 1.2 or the dedicated Muse Voice Transcribe model. It is worth noting because the newest checkpoint is the weaker choice here — an unusual direction, and easy to miss when "latest" normally means "better at everything".
Q · 05 What does it cost to add web search? +
Search grounding is billed separately at $2.50 per 1,000 queries, on top of the token cost of whatever the search returns into your context. Muse Image's built-in search is different — that is included in its flat per-image price.