META MODEL API1M CONTEXTPUBLIC PREVIEWAGENTIC + MULTIMODAL
Muse Spark 1.1 API Pricing
Muse Spark 1.1 is the first Meta model in this catalogue with a metered per-token rate. Meta Superintelligence Labs shipped it on July 9, 2026 alongside the Meta Model API, which publishes a single rate card: $1.25/M input, $0.15/M cached input and $4.25/M output. The API is still labelled a public preview, so treat these rates as pre-GA.
Input - per 1M tokens
$1.25/M
Source dev.meta.ai 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 pre-loaded with the current Meta Model API rates. Tweak the workload split, then share the URL to share the calculation.
$ /mo
Workload split
Prompt cache hit rate
Tokens you can process
—
Words equivalent (English)
—
Effective rate
—
§ 02 / SCENARIOS
Real-world presets.
CODING AGENT
Codebase-scale migration
$0.084/task
LONG DOC
Reading 100-page contracts
$0.097/doc
RAG SUPPORT
Ticket triage
$0.00703/ticket
ASSISTANT
Research planning turn
$0.015/turn
§ 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 546
Words 89
Tokens (estimated) 142 tokens
Cost as input · uncached $0.00018 USD
Cost as output · uncached $0.0006 USD
Cost as cached input $0.00002 USD
| Model | Input /M | Output /M | Effective blended | Context | Best for |
|---|---|---|---|---|---|
| Muse Spark 1.1 Current | $1.25 cache $0.15 | $4.25 | $0.66 agentic 92/8 | 1M | Agentic tool and computer use on Meta's API |
| Gemini 3.5 Flash-Lite | $0.30 cache $0.03 | $2.50 | $0.27 budget 1M peer | 1M | High-volume 1M-context work on a budget |
| Grok 4.3 | $1.25 cache $0.20 | $2.50 | $0.56 same input, cheaper output | 1M | Output-heavy 1M-context agents |
| GPT-5.6 Luna | $1.00 cache $0.10 | $6.00 | $0.76 OpenAI 1M peer | 1.05M | OpenAI-native agent stacks |
| Gemini 3.6 Flash | $1.50 cache $0.15 | $7.50 | $0.96 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.
Practical Muse Spark 1.1 pricing questions, with the published rate card separated from workload assumptions.
Q · 01 What does Muse Spark 1.1 cost on the Meta Model API? +
Meta publishes
$1.25 per 1M input tokens, $0.15 per 1M cached input tokens and $4.25 per 1M output tokens. The rate card is published for the API as a whole rather than per model, and muse-spark-1.1 is the only model ID the models page lists today.Q · 02 Is the Meta Model API generally available? +
No. Meta launched it as a public preview on July 9, 2026, alongside the model. We record the model as preview status for that reason - a preview rate card can move at GA, so treat these numbers as current-but-provisional and re-check before committing a budget.
Q · 03 How much does prompt caching save? +
Cached input is
$0.15/M against a $1.25/M standard input rate, an 88% discount. Meta publishes no separate cache-write charge, so a cache miss bills at the normal input rate. Our effective blend assumes 82% of input tokens hit cache.Q · 04 Does web search grounding cost extra? +
Yes. Web search grounding bills at
$2.50 per 1,000 search queries in addition to the tokens for the request. That is not included in any figure on this page - the calculator and scenarios model token cost only.Q · 05 What are the rate limits? +
Per team, the free tier allows
60 requests per minute and 2,000,000 tokens per minute; the paid tier allows 3,000 requests per minute and 4,000,000 tokens per minute.Q · 06 Is there a batch or priority discount? +
Meta publishes no batch, flex or priority rate for this API today. We do not assume a discount that the vendor has not printed - plan around the standard card until Meta publishes one.
Q · 07 How does the price compare to other 1M-context models? +
At an effective
$0.66/M under our agentic blend it undercuts Gemini 3.6 Flash by 46% and GPT-5.6 Luna by 15%. But Grok 4.3 charges the same $1.25/M input with a $2.50/M output rate and lands 15% cheaper on the same blend, so output-heavy workloads are not automatically better off here.Q · 08 Does Meta publish a maximum output length? +
Not on the models page. The context window is documented at
1,048,576 tokens, but no max output token count is published, so we do not record one. Meta also states that the small steering context it injects into every prompt is not billed.