Build with the API
Visit Meta’s developer portal, check regional eligibility, create a project and generate an API key. Select muse-wonder-1.0.
Open the official quickstartTHE MUSE WONDER FIELD GUIDE
What is Muse Wonder? Meta’s model family for coding and long-running AI agents. Here’s what it costs, where to use it, and how it compares.
Official source·Last verified
Wonder has an API. Wonder Full needs a Muse Code subscription. They are different ways to access different models.
PRICING & AVAILABILITY
API usage, a coding subscription, and the consumer app are separate products. Here’s the distinction.
Official source·Last verified
| Access channel | Muse WonderAPI + CODE | Muse Wonder FullCODE ONLY |
|---|---|---|
| Meta Model APIPay as you go · USD per 1M tokens | Input$4Cached input$0.20Output$20 | No public APINo per-token price published. |
| Muse CodeCoding tool · subscription or API billing | Available on every planAlso available with pay-as-you-go billing. | Power $50 / Maximum $200Per month. Usage limits apply; not unlimited API access. |
| Consumer appMuse app · rollout announced | Coming to subscribers firstAnnouncement describes a future rollout, not confirmed access for every account. | Not separately confirmedNo standalone Full consumer price verified. |
Prices are a verified snapshot, not an offer from this website. Taxes, usage caps, region and eligibility may affect access. Check Meta before paying. Source ↗
MODEL COMPARISON
An independent benchmark snapshot, with the exact model and reasoning setting shown. A score is a starting point, not a verdict on your workload.
Independent evaluation·Last verified
| Model / reasoning | AA Intelligence Index | Input / 1M | Output / 1M | Public API | Source |
|---|---|---|---|---|---|
| Muse Wondermax | 52 | $4 | $20 | Yes | AA |
| Muse Wonder Fullmax | 56 | — | — | No | AA |
| Claude Opus 5.5max, default fallback | 58 | $4 | $20 | Yes | AA |
| GPT-6 Astramax | 53 | $10 | $50 | Yes | AA |
| Gemini 4 Argonhigh | 53 | $2 | $10 | Yes | AA |
Scores: Artificial Analysis Intelligence Index v4.3.2 (higher is better). Prices: evaluator-listed USD per million tokens; rates and tiers can vary. Different reasoning settings are shown explicitly. No tests were run by this website.
GET STARTED
This is an information guide. Create your account and manage model access on the provider’s own website.
Official source·Last verified
Visit Meta’s developer portal, check regional eligibility, create a project and generate an API key. Select muse-wonder-1.0.
Open the official quickstartUse Muse Code with a Power Usage or Maximum Usage subscription. Increasing API reasoning effort does not switch Wonder to Full.
Check model availabilityWatch the official rollout announcement. Consumer access and pricing should not be inferred from developer API or Muse Code plans.
Read the announcementimport os
from openai import OpenAI
client = OpenAI(
base_url="https://api.meta.ai/v1",
api_key=os.environ["MODEL_API_KEY"],
)
response = client.responses.create(
model="muse-wonder-1.0",
input="Explain this codebase's architecture.",
)
print(response.output_text)COMMON QUESTIONS
Official source·Last verified
Start at Meta’s developer portal for API access, or use Muse Code for the coding product. Availability depends on your region, account and current eligibility. This website cannot grant access.
No downloadable Wonder or Wonder Full weights have been identified in the official release materials checked on October 14, 2026. Do not confuse the Wonder family with Meta’s open-weight Muse Glimmer.
Wonder is available through the Model API and Muse Code. Full is the higher-capability variant available through Muse Code’s Power Usage and Maximum Usage subscriptions, without a public per-token API offering.
Wonder targets more complex coding and sustained agent work. Spark is a separate family positioned for lower-cost workloads. Meta’s documentation uses the same request format; the model ID selects the family.
No. This is an independent information and comparison website, with no affiliation with Meta. Provider links lead to their own websites. Any website account belongs only to this guide.
The comparison table uses Artificial Analysis’s own model pages. Meta’s model card is linked separately and documents its evaluation sources and methods. This website has not independently tested these models.
Keep up with Wonder.
New releases, pricing changes and access updates. Get the news in your inbox.
EDITORIAL NOTES
Checked against first-party documentation and the independent evaluator linked in each comparison row. Dates describe editorial checks; information is not updated automatically.
Official source·Last verified