Muse Wonder Price: Full Token and Plan Breakdown
A complete breakdown of Muse Wonder API pricing, subscription plans, and how it compares to frontier models like Claude Opus 5.5.
👁 114 reads
Meta announced the Muse Wonder series on October 12, 2026, positioning it as the successor to Muse Spark at the top of the Meta model line. For developers accessing the API under model ID muse-wonder-1.0 or via OpenRouter as meta/muse-wonder, the base rates are 4 USD per 1M input tokens and 20 USD per 1M output tokens. When caching prompts, the cached input rate drops to 0.20 USD per 1M tokens. Teams tracking reasoning tokens must note that Meta bills all reasoning tokens as output at the standard 20 USD per 1M token rate. Reasoning effort can be configured from minimal to max, with higher effort directly increasing the volume of billed output tokens.
For consumer and development interface usage, Meta offers access through Muse Code subscription tiers priced at 5, 15, 50, and 200 USD per month. Muse Wonder is included across all four subscription tiers. However, the higher-capability variant named Muse Wonder Full carries no API access and is restricted exclusively to the 50 USD Power Usage and 200 USD Maximum Usage plans. Meta documentation specifies that dialing reasoning effort to max on Muse Wonder does not convert it into Muse Wonder Full, as they remain separate models.
Comparing Muse Wonder to Alternative Market Rates
As of the price snapshot on October 12, 2026 at 19:54:27, the standard API price of Muse Wonder exactly matches Claude Opus 5.5, which also lists at 4 USD per 1M input tokens and 20 USD per 1M output tokens. Both sit significantly lower than higher-tier alternatives such as Claude Fable 5.1 and GPT-6 Astra, each priced at 10 USD per 1M input tokens and 50 USD per 1M output tokens. Conversely, Gemini 4 Argon remains the low-cost option in this peer group at 2 USD per 1M input and 10 USD per 1M output tokens.
A team processing a realistic monthly workload of 50 million input tokens, 20 million cached input tokens, and 10 million output tokens would pay 200 USD for regular input, 4 USD for cached input, and 200 USD for output, totaling 404 USD on Muse Wonder. Running that same workload on GPT-6 Astra or Claude Fable 5.1 would cost 500 USD for inputs and 500 USD for outputs, reaching 1,000 USD without caching. Opting for Gemini 4 Argon would cost 100 USD for inputs and 100 USD for outputs. On the API Standard mode, Meta confirms that prompts and responses are not used to train its models.
More on the model: Muse Wonder hub - prices, benchmarks and FAQ
More on this from QUANTSLANT
guides·Cloudflare WARP and 1.1.1.1: Is It a VPN and Worth Using?Cloudflare WARP encrypts device traffic, but it is not a traditional VPN. See how it works, how it differs from 1.1.1.1, and when it is worth using.