$4 in. $20 out. Full starts at $50/month.
Muse Wonder API pricing is $4 per million uncached input tokens, $0.20 per million cached input tokens, and $20 per million output tokens. Muse Wonder Full is a separate variant included in Muse Code Power Usage ($50/month) and Maximum Usage ($200/month).
Budget by usage
Count the whole agent workflow: repeated context, output and tool activity.
Check the Code plans
Full has no published Meta API token rate. A monthly plan is a different purchase.
Independent editorial guide. These are Meta’s and OpenRouter’s published prices, not VSModels packages. We have not run a paid Muse Wonder test.
Where can you use Muse Wonder?
Start with the access route: an API gateway, a web workspace, or a coding client. The 13 entries below include several routes to the same Meta model; they are not 13 independent inference hosts.
Showing 13 of 13 documented access routes.
On small screens, scroll the table sideways to see prices and sources.
| Platform / route | Model & Full access | Published cost | Evidence & conditions |
|---|---|---|---|
| Meta Model APIDirect API | muse-wonder-1.0Full: Not on Model API | $4 / $0.20 / $20 | Model maker documentation Access notesMeta account and API billing; regional eligibility applies. |
| OpenRouterAPI gateway | meta/muse-wonderFull: No Full API listing verified | $4 / $0.20 / $20 | Platform model/provider listing Access notesThe inspected provider row is Meta. Gateway access is not another model host. |
| Vercel AI GatewayAPI + browser playground | meta/muse-wonder-1.0Full: No Full API listing verified | $4 / $0.20 / $20 | Platform announcement + model listing Access notesVercel key and team billing; no separate Meta account required. Playground uses API rates. |
| WritingmateWeb workspace | Muse WonderFull: No Full listing verified | Writingmate plan allowances | Platform catalog + access statement Access notesIts $4/$20 figures are upstream reference rates, not Writingmate checkout prices. Exact Wonder allowance not verified. |
| Muse CodeFirst-party coding CLI | Wonder; Full on eligible plansFull: Power / Maximum subscriptions | API usage or subscription; Full $50 / $200 monthly | Model maker subscription documentation Access notesUsage limits and default Contributor mode apply. Subscription credential is for Muse Code only. |
| PiClient through OpenRouter | meta/muse-wonderFull: No Full route verified | Underlying provider billing | Client model configuration Access notesRequires your provider key. The inspected Pi entry lists text/image input; do not assume all upstream modalities. |
| OpenCodeCustom Meta provider | Explicit Wonder model selection neededFull: No Full integration verified | Underlying Meta API billing | Meta provider setup; sample caveat Access notesMeta describes Wonder compatibility, but its OpenCode sample still names Spark. Treat this as integration guidance, not a tested Wonder preset. |
| Codex CLICustom Meta provider | muse-wonder-1.0Full: No Full integration verified | Underlying Meta API billing | Meta-published configuration Access notesExternal-provider configuration, not an OpenAI-hosted Muse model or a ChatGPT subscription benefit. |
| Claude CodeCustom Meta Messages provider | muse-wonder-1.0Full: No Full integration verified | Underlying Meta API billing | Meta-published configuration Access notesUses a Meta key and endpoint; model aliases/subagents need matching configuration. Not Anthropic-hosted Muse access. |
| BeatAPIThird-party relay | muse-wonderFull: No Full listing verified | $4 / not verified / $20 | Relay documentation only Access notesRequires a funded account; signup credits do not apply. Cache rate not established by the inspected table. |
| CompanyFabricThird-party gateway + playground | meta/muse-wonderFull: Explicitly excludes Full API | $4 / $0.20 / $20 | Relay listing only Access notesPublisher says it routes to Meta and offers prepaid or BYOK. Identity, fulfillment and billing were not independently tested. |
| Decisions APIOpenRouter relay + playground | meta/muse-wonderFull: No Full listing verified | $5.72 / $0.29 / $28.58 | Relay listing only Access notesPublished platform prices include markup; 8,192 output cap, 40 messages and service-specific upload limits. |
| VelobaseConfigured project gateway | meta/muse-wonderFull: No Full listing verified | $4.40 / not verified / $22 | Gateway listing only Access notesProject pricing is fixed at import/configuration. The $4/$20 provider row is upstream reference pricing, not the displayed customer price. |
“Listing verified” means we read a platform’s own model page or documentation. It does not verify fulfillment, model identity during inference, uptime, account eligibility or a successful payment. All rows share the October 15 check date. Missing prices are unverified, not zero.
Wonder Full has a narrower access path.
Meta documents Full only inside eligible Muse Code subscriptions. A client that can call the standard Wonder API does not inherit Full access or subscription usage. We do not recommend routing Muse Code subscription credentials into other tools.
Meta: model versions and availability Meta: Muse Code subscriptions and usageFor browser use, inspect Writingmate’s actual plan allowance or Vercel’s metered playground. For an app, compare the direct Meta, OpenRouter and Vercel routes first; the separately labeled relay listings show why the upstream rate is not always the checkout rate. Coding clients need a configured provider and key.
Coverage is a public-documentation snapshot, not an exhaustive web census. We found no model-specific first-party confirmation for Poe, Cursor’s built-in catalog, Cline or Kilo during this audit; that is not proof of non-support through custom providers. The consumer Muse rollout remains separate from verified developer access. Sources are linked on every row.
Muse Wonder API pricing, per million tokens.
| Billable component | Price | Budget example |
|---|---|---|
| Uncached input | $4 / 1M tokens | 100,000 tokens → $0.40 |
| Cached input | $0.20 / 1M tokens | 100,000 cached tokens → $0.02 |
| Output | $20 / 1M tokens | 25,000 tokens → $0.50 |
| Web search grounding | $2.50 / 1,000 queries | 100 queries → $0.25, plus tokens |
Meta lists no long-context price premium. Cached-prefix hits get the lower rate; inspect cached_tokens rather than assuming every repeated prompt is cached. Wonder’s API offers Standard mode only, without a Contributor discount.
The exact documented context window is 1,048,576 tokens. Choosing max reasoning on Wonder does not turn it into Wonder Full.
Estimate a workload, including cache hits.
At 1M total input and 250k output tokens, the uncached bill is $9. With half the input served from cache, it becomes $7.10. That is an illustrative calculation, not measured usage.
- Uncached input
- $2.00
- Cached input
- $0.10
- Output
- $5.00
- Search
- $0.00
Cache savings: $1.90 for this assumed workload.
Illustrative aggregate usage, not a single request. Cached tokens are part of total input, never added twice. Use provider-reported billable output, not just visible answer length. Excludes taxes, currency conversion, gateway fees and non-search tool costs. Prices checked Oct 15, 2026; no API call or VSModels credit charge.
Formula: uncached input × $4/1M + cached input × $0.20/1M + output × $20/1M + searches × $2.50/1,000. The cache rate is 95% below the uncached input rate; that is not a 95% discount on the entire bill.
How much does Muse Wonder Full cost?
| Plan | Monthly price | Published usage allowance | Full access |
|---|---|---|---|
| Power Usage | $50 | 20× Everyday Usage | Included in Muse Code |
| Maximum Usage | $200 | 80× Everyday Usage | Included in Muse Code |
These are usage-limited subscriptions, not unlimited tokens. Meta describes Everyday Usage as 10–50 prompts per five hours. The multipliers are plan allowances, not a guaranteed number of Full requests; check your actual onboarding terms and region.
The subscription credential works only in Muse Code. Additional API keys are billed separately. At a plan limit, upgrade or wait for refresh; Maximum users must wait. Subscriptions renew monthly. Meta says to cancel at least 24 hours before the next billing date to avoid renewal.
Subscriptions default to Contributor mode: extra usage in exchange for allowing training on submitted content. You can disable it. Check this setting before sending private code.
Meta and OpenRouter list the same token rates.
| Access path | Input | Cached input | Output |
|---|---|---|---|
| Meta · muse-wonder-1.0 | $4 | $0.20 | $20 |
| OpenRouter · meta/muse-wonder · Meta provider | $4 | $0.20 | $20 |
OpenRouter’s provider table lists Meta at these rates. Equal token prices do not establish equal final bills: check funding fees, taxes, routing and tool billing in the service you use. Our calculator applies Meta’s documented search rate only.
Start with where the work runs.
An app or a scheduled agent
Estimate API usage across every turn. Long outputs can dominate: in our default example, output costs $5 of the $7.10 total.
Coding in Muse Code
Compare a representative week against the plan allowance. Do not treat a $50 subscription as $50 of API credit.
Wonder Full specifically
Check Power or Maximum eligibility. An API reasoning setting cannot substitute for access to the Full variant.
An existing OpenRouter setup
Confirm the exact model ID, provider and billed usage. A listing on OpenRouter does not add the model to VSModels.
There is no universal subscription break-even point here: model access, usage limits and execution environment differ. Compare successful tasks and your actual receipts.
Muse Wonder price, without the ambiguity.
Is Muse Wonder Full available through Meta’s API?
Meta documents Full as Muse Code subscription-only. Its model page leaves API token prices blank. The standard Wonder API model is muse-wonder-1.0.
Does $50/month include unrestricted API calls?
No. The listed Power plan covers Muse Code usage. Extra API keys use separate pay-as-you-go billing.
Is the cached-input discount applied to output?
No. It applies only to input tokens served from cache. Output retains its separate rate.
Is this the price of the consumer Muse app?
This guide covers developer API usage and Muse Code. Meta’s launch announcement describes a separate rollout into the consumer Muse experience; do not assume the products share a subscription.
Meta: October 12 Wonder launch announcementCan I use Muse Wonder on VSModels?
This page is editorial coverage, not an integration announcement. It does not add Muse Wonder or Full to the executable model list.
Published rates. Explicit assumptions.
Checked October 15, 2026. Official documents establish rates and access; OpenRouter corroborates its own listing. Calculator results are our arithmetic, not invoices or benchmark runs. This page is a dated snapshot, not a live price feed. Confirm current regional availability, taxes and terms with the provider before purchase.
- Meta: API pricing and rate limits
- Meta: Muse Wonder and Full model page
- Meta: model versions and availability
- Meta: Muse Code subscriptions and usage
- OpenRouter: Muse Wonder provider pricing
- Meta: October 12 Wonder launch announcement
- Vercel: Muse Wonder API and playground listing
- Vercel: October 12 AI Gateway announcement
- Writingmate: Muse Wonder access and plan allowances
- Pi: Muse Wonder through OpenRouter
- Meta: external coding-agent configurations
- BeatAPI: Muse model API documentation
- CompanyFabric: Muse Wonder API listing
- Decisions API: Muse Wonder relay and playground
- Velobase: Muse Wonder gateway listing