Meta's muse-wonder-1.0 costs $4 per million input tokens and $20 per million output tokens on the Meta Model API, with cached input at $0.20. DeepSeek V4.1 Flash and GLM-5.3 list lower per-token prices. Muse Wonder Full is a separate subscription-access model with no published API token rate to put in this table. Compare observed task costs and results before drawing a value conclusion.
What did Meta release with Muse Wonder on October 12?
Meta announced the Muse Wonder series on October 12, 2026. It introduced Muse Wonder and Muse Wonder Full as two models for personal and professional agent tasks. Muse Wonder became available through the Meta Model API and Muse Code; Wonder Full was offered to eligible Muse Code subscribers. Meta says Muse Wonder will reach the consumer Muse experience later, starting with subscribers.
That access split matters for price comparisons. Developers can buy metered API usage of Muse Wonder 1.0. Meta's model page lists Wonder Full under Muse Code subscription access and does not publish a comparable API input or output rate for it. We therefore compare the API model that a developer can actually price today.
What does Muse Wonder add for long-running agents?
Meta positions Muse Wonder around complex coding, computer use and tasks that stretch across many steps. Its model page lists a one-million-token context window. Meta calls two of its design goals Active Memory, carrying relevant background across conversations, and Active Focus, staying on a difficult task while trying alternatives, handling obstacles and asking for clarification when necessary.
Meta reports results for professional work, memory, agent use and sustained coding. These are provider-published evaluations and demonstrations, not head-to-head tests run by SaveMyToken against DeepSeek or GLM. The practical question for a team is whether the longer-running behavior improves its own accepted output enough to justify the higher API rate.
What is the price of Muse Wonder Full?
There is no public per-million-token Meta Model API price for muse-wonder-full-1.0 in Meta's current model and pricing pages. It is available through qualifying Muse Code subscription plans. We do not convert a monthly subscription into an invented token price or put zero in the table.
How should you compare Muse Wonder with DeepSeek and GLM?
Use the same task and record uncached input, cache reads, generated output, retries and tool calls. For a token-only example, one million uncached input tokens plus 200,000 output tokens costs $8 at Muse Wonder's listed Standard rate, $2.28 on GLM-5.3, or $0.27–$0.54 on DeepSeek V4.1 Flash, depending on its off-peak or peak window. These are rate calculations, not evidence that the models produce equally useful results.
Meta charges $0.20 per million cached input tokens when a prefix qualifies and the response reports a cache hit. Its Model API web-search tool also has a separate per-query charge. DeepSeek's deepseek-flash endpoint currently serves V4.1 Flash, with lower off-peak prices outside its published weekday UTC peak windows. GLM-5.3 is shown at Z.ai's direct USD API rate.
How do official API prices compare in October 2026?
USD per one million tokens, checked October 14, 2026. Muse Wonder comes first, followed by DeepSeek and GLM-5.3. The remaining models are reference points. A cached-input price applies only to eligible cache hits; it is not the default input price.
| Model / provider | Input | Output | Cached input |
|---|---|---|---|
| Muse Wonder 1.0 · Meta | $4.00 | $20.00 | $0.20 |
| DeepSeek V4.1 Flash | $0.15–$0.30 | $0.60–$1.20 | $0.003–$0.006 |
| GLM-5.3 · Z.ai | $1.40 | $4.40 | $0.26 |
| Gemini 4 Argon · Google | $2.00 | $10.00 | $0.10 |
| Claude Opus 5.5 · Anthropic | $4.00 | $20.00 | $0.20 |
| Claude Sonnet 5.5 · Anthropic | $2.00 | $10.00 | $0.10 |
| GPT-6 Astra · OpenAI | $10.00 | $50.00 | $1.00 |
| GPT-6.1 Sol · OpenAI | $2.00 | $10.00 | $0.10 |
| MiMo V2.6 Pro · Xiaomi | $0.435 | $0.87 | $0.0036 |
| Kimi K3 · Moonshot | $3.00 | $15.00 | $0.30 |
| Mistral Large 4 Preview · Mistral AI (France) | $0.68 | $2.09 | $0.07 |
Want to try a cost-effective Coding Plan featuring DeepSeek V4.1 Flash, GLM-5.3, MiMo V2.6 Pro, and Kimi K3? Visit our coding plan.
- Meta:
muse-wonder-1.0Standard API pricing. Wonder Full has no published metered API rate and is not silently assigned this row's price. - DeepSeek: ranges show off-peak first, peak second; peak is 01:00–04:00 and 06:00–10:00 UTC on weekdays, excluding Chinese public holidays.
- Gemini: introductory Standard rates through November 30; input/output/cache become $4/$20/$0.20 on December 1. The cache rate is for implicit caching.
- Claude, OpenAI and Kimi: cache writes may be charged separately. OpenAI's listed rates use Standard short-context processing; other tiers and tool charges are outside this table.
- MiMo: Xiaomi's overseas real-time Pro rate. Domestic CNY, batch and UltraSpeed rates differ.
- Mistral: Large 4 public-preview launch discount is 50% for two weeks from October 6; regular listed input/output/cache prices are $1.36/$4.18/$0.14. Check the active rate before purchasing.
Sources and price basis
- Meta's October 12 research announcement ↗ — release, access split and Active Memory/Focus descriptions.
Official API rates were checked October 14, 2026. Rates can change. No cross-model quality or task-cost benchmark was run for this article.