🔍 Read the full analysis: A Look At Muse Wonder Full’s Move Into The AI Frontier on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Meta has released Muse Wonder and Muse Wonder Full, two proprietary models with 1-million-token context windows. Artificial Analysis scored Muse Wonder Full at 56 on its Intelligence Index, while the API-available Muse Wonder scored 52; the evaluator estimated $4.52 per benchmark task for Muse Wonder, substantially more than some models with similar scores.
Meta has released Muse Wonder and Muse Wonder Full, two proprietary AI models that Artificial Analysis evaluated on October 12, 2026. The evaluator gave Muse Wonder Full an Intelligence Index score of 56, while the version available through Meta’s API scored 52; its estimated cost per benchmark task was $4.52, highlighting a pricing gap against models with similar scores.
Both models have a 1-million-token context window and accept text, images, speech or audio, and video as input, returning text. Meta makes Muse Wonder available through its API and Muse Code, its coding agent. The listed API rates are $4 per million input tokens, $0.20 per million cached input tokens, and $20 per million output tokens. Muse Wonder Full is available only through Muse Code and has no published per-token price.
Artificial Analysis reported that Muse Wonder Full scored 56 on its Intelligence Index v4.3, eight points above Muse Spark 1.3’s 48. The evaluator said the model improved across all ten evaluations in the index and placed in the top three on six. Its reported Terminal-Bench 4.0 result was 57%, compared with 33% for Muse Spark 1.3. On GDPval-AA, which tests economically useful knowledge work, it scored 1,857 Elo, close to the reported scores for Claude Opus 5.5 and Claude Sonnet 5.5.
Muse Wonder scored 52, matching GPT-6.1 Sol’s reported index score. Artificial Analysis estimated that one Intelligence Index task cost $4.52 with Muse Wonder, compared with $0.72 for Sol, $1.99 for Gemini 4 Argon, and $3.26 for GPT-6 Astra. These are evaluator estimates for the benchmark workload, not a universal charge for every use. The source attributes much of Muse Wonder’s task cost to input: it processed 7.69 million input tokens per task, with input accounting for $3.43, or 76%, of the estimated total.
Infographic 01 · Where Meta’s new pair lands
Muse Wonder Full ties for second. Muse Wonder ties GPT-6.1 Sol.
Artificial Analysis scores 27 frontier and open-weight configurations on ten evaluations. Meta’s two new models land near the top of the list. Six weeks earlier, Muse Spark 1.3 scored 48.
Intelligence Index score, 27 configurations
Higher is better. Reasoning effort in brackets, as Artificial Analysis tested it.
Availability: Muse Wonder runs on Meta’s API and in Muse Code. Muse Wonder Full is only available to Muse Code users and has no per-token price.
Source: Artificial Analysis, Muse Wonder analysis (12 Oct 2026) and model leaderboard, read 13 Oct 2026. Intelligence Index v4.3, ten evaluations.
Powered by Thorsten Meyer AI
Strong Scores Meet High API Costs
The release gives developers two different propositions: a stronger model reserved for Muse Code, and a model that can be called through Meta’s API at published token rates. The split matters for buyers comparing performance and cost. Muse Wonder Full’s reported results put it near leading models on several measures, but its lack of a per-token price means users outside Meta’s coding tool cannot directly compare its task-level economics from the information provided.
For the purchasable API model, the evaluator’s cost estimate complicates the performance result. Muse Wonder matched Sol’s reported score of 52 but had a much higher estimated task cost in this evaluation. That difference could affect adoption by developers running large workloads, where token consumption and repeated tasks influence operating budgets. The estimate should not be treated as a price guarantee: actual spending depends on workload, token use, caching, and application design.
The benchmarks also point to practical gains for coding and automation. Artificial Analysis reported Muse Wonder Full at 57% on Terminal-Bench, while its Coding Agents Index gave Muse Code plus Wonder Full a score of 68, tied with Claude Code plus Sonnet 5.5 in that comparison. The agent score combines the model and its software harness; it does not establish that the model would get the same result in another coding environment.
As an affiliate, we earn on qualifying purchases.
Six Weeks After Muse Spark
The release comes six weeks after Muse Spark 1.3, according to the source material. Artificial Analysis’s comparison places the new models alongside systems from Anthropic, Google, and OpenAI using its own index and task-cost methodology. These scores are evaluator findings, not claims that every model will perform the same way across different products or user workloads.
The source reports that Muse Wonder’s API rates are $4 per million input tokens and $20 per million output tokens, the same listed rates as Claude Opus 5.5. It says several other compared models, including Claude Sonnet 5.5, Gemini 4 Argon, and GPT-6.1 Sol, list rates of $2 per million input and $10 per million output. Muse Spark 1.3 was listed at $1.25 per million input tokens and $4.25 per million output tokens. The task-cost estimates reflect benchmark token use as well as per-token rates, so they are not interchangeable with list prices.
Artificial Analysis also reported hallucination rates of 23% for Muse Wonder Full and 27% for Muse Wonder, down from 33% for Muse Spark 1.3 in the supplied comparison. The evaluator’s AA-Omniscience results were 30 and 25, respectively. Those results suggest improvement over the predecessor in the stated tests, while leaving a measurable risk of incorrect answers. They do not establish error rates for every application or real-world deployment.
As an affiliate, we earn on qualifying purchases.
Questions About Cost and Real-World Results
The supplied material does not include a public per-token price for Muse Wonder Full, which is limited to Muse Code. Without that information, outside users cannot calculate a directly comparable API cost for the higher-scoring model. It is also unclear whether Meta plans to make it available through an API or publish broader pricing details.
The $4.52 task estimate is tied to Artificial Analysis’s benchmark workload and reported token use. It may not predict the cost of a customer’s own application, where prompts, caching, context length, and output requirements can differ. The source also does not provide independent replication or confidence intervals for every reported comparison, so small score differences should not be treated as definitive rankings.
Benchmark improvements do not remove the need to check outputs. The reported hallucination rates remain above zero, and the source does not establish how either model performs across specific industries, languages, or production conditions. Meta’s own statements about intended use and deployment safeguards are not included in the material.
AI input and output token counters
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Pricing and Access Will Shape Adoption
Developers can test Muse Wonder through Meta’s API or use it in Muse Code, while Muse Wonder Full remains limited to Muse Code based on the information provided. Meta’s next relevant step for prospective API customers would be to clarify whether the Full model will become available outside that product and, if so, how it will be priced.
Users evaluating Muse Wonder can compare their own token consumption and task results with the benchmark estimates, rather than assuming the reported $4.52 applies to every job. Further independent testing would help show whether the models’ reported gains hold across different workloads and whether their costs remain competitive in practice. The source material does not give a timeline for additional releases, pricing changes, or wider access.
As an affiliate, we earn on qualifying purchases.
Key Questions
What are Muse Wonder and Muse Wonder Full?
They are two proprietary Meta AI models released after Muse Spark 1.3. Both have a reported 1-million-token context window and accept text, images, audio or speech, and video as input.
Which model can developers access through Meta’s API?
Muse Wonder is available through Meta’s API and Muse Code. Muse Wonder Full is available only through Muse Code according to the source material.
How did the models score in Artificial Analysis’s evaluation?
Artificial Analysis gave Muse Wonder Full a score of 56 and Muse Wonder a score of 52 on its Intelligence Index v4.3. The results are specific to the evaluator’s tests and do not guarantee performance on every task.
Why is Muse Wonder’s estimated task cost drawing attention?
Artificial Analysis estimated a cost of $4.52 per Intelligence Index task for Muse Wonder, compared with $0.72 for GPT-6.1 Sol, which had the same reported index score. The estimate reflects the benchmark’s token use and is not a fixed cost for all applications.
Is Muse Wonder Full available at a published API price?
No per-token price for Muse Wonder Full is given in the source material. It is described as available only through Muse Code, so a directly comparable API task cost is not available there.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
