🔍 Read the full analysis: Muse Wonder Full Joins The AI Frontier As Muse Wonder Charges Like It Did Too on ThorstenMeyerAI.com
Get privacy and security gear delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Meta has released Muse Wonder and Muse Wonder Full, six weeks after Muse Spark 1.3. Artificial Analysis gave Wonder Full a score of 56, while the API-available Wonder scored 52; one benchmark task with Wonder costs $4.52, compared with $0.72 for GPT-6.1 Sol at the same score.
Meta has released Muse Wonder and Muse Wonder Full, two proprietary AI models that scored 52 and 56, respectively, on Artificial Analysis’s Intelligence Index v4.3. The stronger model is available only through Meta’s Muse Code product, while the API-accessible Wonder costs $4.52 per benchmark task in the evaluator’s comparison—far more than another model with the same score.
Artificial Analysis published its evaluation on October 12, 2026. Wonder Full scored eight points above Muse Spark 1.3, which received 48, and improved across all ten evaluations in the index. Among 27 configurations compared by the source, only Claude Opus 5.5 at its maximum setting scored higher, with 58; Claude Sonnet 5.5 at its maximum tied Wonder Full at 56. These are benchmark results, not guarantees of performance in every use case.
The two models share a one-million-token context window and accept text, images, audio and video as input, returning text. Meta offers Muse Wonder through its API and inside Muse Code. Its list price is $4 per million input tokens, $0.20 per million cached input tokens and $20 per million output tokens. Wonder Full is available only to Muse Code users and has no published per-token price in the source material, making its comparable API cost unknown.
Wonder Full’s reported strengths include a 57% Terminal-Bench 4.0 score, up from 33% for Spark 1.3, and a 1,857 Elo rating on GDPval-AA v2.1, which measures economically useful knowledge work. It ranked among the top three on six of the ten evaluations. In Artificial Analysis’s Coding Agents Index, Muse Code paired with Wonder Full scored 68, tying the listed Claude Code and Sonnet 5.5 configuration. That result evaluates the model and its software harness together, not the model alone.
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
Benchmark Gains Meet API Costs
The release gives developers and businesses a new option from Meta, but the two models serve different access and pricing needs. Wonder Full’s benchmark performance puts Meta’s Muse Code system alongside leading coding-agent configurations in this evaluation. At the same time, the stronger model’s limited availability means that developers cannot choose it through the API based on the information provided.
For customers comparing API models, the central issue is the reported cost per task. Artificial Analysis estimated $4.52 for Muse Wonder on one Intelligence Index task. GPT-6.1 Sol scored the same 52 for $0.72 in the same comparison; Gemini 4 Argon scored 53 for $1.99. Such estimates depend on the benchmark’s task and token use, so they should not be treated as a universal price for every real-world request. Still, they indicate that headline model scores alone may not tell buyers what a workload will cost.
Artificial Analysis’s breakdown attributes 76% of Wonder’s task cost to input tokens. It reports that Wonder processes 7.69 million input tokens per benchmark task, compared with 6.03 million for Spark 1.3. That makes input volume a major factor in the evaluator’s cost estimate, alongside Meta’s list prices. Organizations considering the model may need to test their own prompts and usage patterns before estimating budget impact.
As an affiliate, we earn on qualifying purchases.
From Spark 1.3 to Wonder
The release follows Muse Spark 1.3 by six weeks, according to the source report. In that interval, the reported Intelligence Index score rose from 48 to 56 for Wonder Full, while the API model Wonder scored 52. The source describes improvement across all ten index evaluations for Wonder Full, but those scores reflect Artificial Analysis’s particular testing framework.
Meta’s published pricing for Wonder is higher than the reported list prices for several comparison models. The source gives Spark 1.3’s rates as $1.25 per million input tokens and $4.25 per million output tokens; Wonder is listed at $4 and $20. It also reports $2 per million input tokens and $10 per million output tokens for Sonnet 5.5, Gemini 4 Argon and GPT-6.1 Sol. The task-cost comparisons combine model pricing with benchmark token consumption, rather than comparing list prices alone.
Scores on individual measures vary. Wonder Full received 30 on AA-Omniscience, while Wonder received 25. The source reports hallucination rates of 23% and 27%, respectively, compared with 33% for Spark 1.3. Among the listed models scoring at least 45 on the Intelligence Index, Gemini 4 Argon’s reported rate was 15%. These figures describe results under the evaluator’s tests; they do not establish how often a model will make errors in every setting.
“The capability gap closed. The price gap did not.”
— Thorsten Meyer, ThorstenMeyerAI.com
As an affiliate, we earn on qualifying purchases.
Access, Costs and Error Rates
The source does not provide a per-token price for Muse Wonder Full, which is restricted to Muse Code. Its task-level cost therefore cannot be compared directly with API models using the figures supplied. It is also unclear whether Meta plans to offer Full through an API or publish additional pricing.
The $4.52 estimate for Wonder reflects Artificial Analysis’s benchmark task and token consumption. Actual costs may differ across prompts, applications and usage patterns. The evaluation also reports hallucination rates, but those measurements do not determine a model’s accuracy in a particular customer workflow. The source does not give uncertainty ranges for every result, so small score differences should not automatically be read as decisive performance gaps.
Meta’s rationale for Wonder’s pricing, and whether it will change, is not included in the source material. Nor does the report establish how the models perform on independent tasks beyond the evaluations described.
As an affiliate, we earn on qualifying purchases.
More Pricing and Usage Detail
The immediate next step for prospective users is to compare the models on their own workloads, including both output quality and input-token consumption. API customers can access Muse Wonder at Meta’s published rates, while Wonder Full remains tied to Muse Code based on the available information.
Further clarity would depend on whether Meta publishes additional details about Wonder Full’s availability and cost, and whether it changes Wonder’s API pricing. More evaluations could also show how the reported benchmark gains translate to other applications. No future release dates or pricing changes are specified in the source.
AI model performance benchmarking tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What are Meta’s new Muse models?
Muse Wonder and Muse Wonder Full are proprietary models released by Meta. Both accept text, images, audio and video, have a one-million-token context window, and return text.
Which model can developers access through Meta’s API?
Muse Wonder is available through Meta’s API and inside Muse Code. Wonder Full is available only to Muse Code users, according to the source.
How did the models score?
Artificial Analysis gave Wonder Full a score of 56 and Wonder a score of 52 on Intelligence Index v4.3. The results apply to that evaluator’s benchmark and do not predict performance in every task.
Why is Muse Wonder’s task cost drawing attention?
Artificial Analysis estimated a cost of $4.52 per Intelligence Index task for Wonder, compared with $0.72 for GPT-6.1 Sol, which scored 52 in the same comparison. The estimate depends on the benchmark workload and token use.
How much does Muse Wonder Full cost?
The source gives no per-token price for Wonder Full. It says the model is available only through Muse Code, so a comparable API task-cost figure is not available in the report.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
