Meta rows use official Meta Model API pay-as-you-go prices in Standard mode, where your content is not used for training. Cached input applies to cached prompt prefixes, there is no long-context premium, and the discounted Contributor mode is noted but not used as the default rate.
Choose a current Meta model from the same pricing rows, enter input tokens, output tokens, and request volume, then review the estimated monthly API cost.
Start with realistic input tokens, output tokens, model choice, request volume, and any cached prompt-prefix pricing rules that apply. Then compare the estimate against current model pricing and budget limits before launch.
These rows use Meta Model API pay-as-you-go prices in Standard mode. Muse Wonder is in preview, and availability varies by country under Meta's Geographic Use Policy.
In Standard mode your content is not used to train Meta models. Contributor mode (muse-spark-1.3-contributor) costs $0.10 input and $0.20 output per 1M tokens in exchange for training use, so it is noted here but never used as the default rate. Muse Wonder has no Contributor mode on Model API.
Cached input applies to tokens served from a cached prompt prefix; check cached_tokens in the response. Meta charges no long-context premium, so the same rate applies up to the 1,048,576-token context window.
Muse Spark 1.2 and 1.1 cost the same as 1.3 and are not listed separately. Web search grounding adds $2.50 per 1,000 queries. No Batch rate is published.
Open a model page for its verified source, workload examples, price history, lifecycle details, and nearby alternatives.
Null or unverified values are displayed as Not listed, never as zero.
All prices in USD per 1M tokens.
| Model | Input | Output | Cached Input | Verified |
|---|---|---|---|---|
Muse Spark 1.3latestModel ID: muse-spark-1.3DetailsProvider: MetaPrice shown: Standard | Input · USD per 1M input tokens$1.25 | Output · USD per 1M output tokens$4.25 | Cached input · USD per 1M cached input tokens$0.15 | Evidence |
Muse Wonder 1.0previewModel ID: muse-wonder-1.0DetailsProvider: MetaPrice shown: Standard | Input · USD per 1M input tokens$4.00 | Output · USD per 1M output tokens$20.00 | Cached input · USD per 1M cached input tokens$0.20 | Evidence |
Use these Meta pricing rows as planning inputs: calculate your monthly API cost with realistic token counts, or compare this provider with other AI models before routing production traffic.
Meta API pricing is the usage-based cost of running Muse models through API requests. The practical estimate depends on the selected model, input tokens, output tokens, monthly request volume, retries, and any listed caching or discount rules.
Estimate one representative request first: input tokens, expected output tokens, selected Muse models, request volume, retry rate, and cached prompt-prefix pricing assumptions when they are listed. Then multiply by monthly traffic.
Open the Meta API cost calculatorUse the table on this page to compare available Meta rows per 1M tokens. Not listed values stay visible as Not listed, so unknown prices are not accidentally treated as zero.
Compare total workflow cost, not only the cheapest input token price. Output length, quality, retries, context size, and fallback routing can change the real monthly API bill.
Compare Meta with other providersUse cases are planning categories, not pricing promises. Estimate each workflow with its own token shape, expected answer length, quality requirement, and fallback behavior.
Muse Wonder targets multi-file coding changes and long agent runs; compare its rate with frontier models once retries and tool calls are included.
Muse Spark 1.3 is Meta's lower-cost model for tool loops, coding assistants, and long-context reasoning, with no long-context premium up to its 1M context window.
Contributor mode cuts Muse Spark token prices sharply in exchange for letting Meta train on your prompts and responses; decide whether that is acceptable before comparing costs.
Compare Meta with OpenAI, Claude, Gemini, and Qwen by input price, output price, cached-input behavior, context needs, latency, quality, and cost per successful task. Always verify current provider pricing before launch or procurement decisions.
Model a normal, high, and worst-case month with expected requests, input tokens, output tokens, retries, and model routes before exposing a feature to real traffic.
Decide which app, workspace, customer tier, or internal team can spend against the provider budget. Review burn rate weekly during launches and pricing changes.
Shorten repeated instructions, cap output length, avoid sending full history when a summary works, and route routine tasks to lower-cost models when quality testing supports it.
The existing calculator route is linked directly without invented provider or model query parameters.
Pricing data is informational and may lag provider changes. Always verify final billing decisions against provider pricing pages, contracts, dashboards, and invoices.
Meta rows use official Meta Model API pay-as-you-go prices in Standard mode, where your content is not used for training. Cached input applies to cached prompt prefixes, there is no long-context premium, and the discounted Contributor mode is noted but not used as the default rate.
Meta API cost depends on the selected model, input tokens, output tokens, request volume, retries, and any cached prompt-prefix pricing or discount rules that apply. Use the pricing table and calculator with your own workload assumptions instead of relying on a single average price.
Estimate one request with input tokens, expected output tokens, model choice, and current per-1M-token pricing. Then multiply by requests per user, active users, days per month, retry rate, and fallback calls.
Muse models are usually evaluated for workflows such as long-running agents and coding, cost-aware agentic work, data-use trade-offs. The best route depends on quality, latency, context size, output length, and monthly budget.
Meta can be cheaper or more expensive depending on the exact model, prompt size, output length, cache eligibility, retry rate, and quality target. Compare total workflow cost across providers before switching production traffic.
Not listed means the current local pricing source does not expose a confirmed numeric value for that field. It is intentionally not treated as zero. Check the official Meta pricing page and billing dashboard before launch.
Control spend by setting monthly budget limits, measuring input and output tokens separately, capping output length, reducing repeated context, reviewing retry cost, and comparing model routes before high-volume traffic grows.