🔍 Read the full analysis: Muse Wonder Full Takes On The AI Frontier As Muse Wonder Charges on ThorstenMeyerAI.com
Get the latest gadgets 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, two proprietary models that Artificial Analysis evaluated on October 12, 2026. Muse Wonder Full scored 56 on the Intelligence Index but is limited to Muse Code; the API model scored 52, with an estimated benchmark-task cost of $4.52.
Meta has released Muse Wonder and Muse Wonder Full, two proprietary AI models that scored 52 and 56, respectively, on the Artificial Analysis Intelligence Index v4.3. The stronger model is available only inside Meta’s Muse Code coding agent, while the API-available Muse Wonder costs an estimated $4.52 per benchmark task—well above several models with similar index scores, according to the evaluator’s results.
Artificial Analysis published its evaluation on October 12, 2026, six weeks after Meta released Muse Spark 1.3. It gave Muse Wonder Full a score of 56, up from 48 for Spark 1.3, and reported improvements across all ten evaluations in the index. Muse Wonder scored 52. The evaluator’s comparison puts Muse Wonder Full behind Claude Opus 5.5 at 58 and level with Claude Sonnet 5.5 at 56; Muse Wonder’s score matches GPT-6.1 Sol.
The two models share a one-million-token context window and accept text, images, audio and video, returning text. Their availability differs. Meta offers Muse Wonder through its API and in Muse Code, at list prices of $4 per million input tokens and $20 per million output tokens, with cached input listed at $0.20 per million tokens. Muse Wonder Full has no per-token price and is available only to Muse Code users, so its API cost cannot be compared from the supplied information.
Artificial Analysis reported particularly strong Muse Wonder Full results on several individual tests: it ranked first on GDP.pdf and CritPt, and third on Terminal-Bench 4.0, AA-Briefcase, AutomationBench-AA and SciCode. Its Terminal-Bench score was 57%, compared with 33% for Muse Spark 1.3. In the evaluator’s Coding Agents Index, Muse Code paired with Wonder Full scored 68, tying the Claude Code and Claude Sonnet 5.5 configuration. That result measures the combined model and coding-agent system, 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
Strong Scores Meet Higher API Costs
The release presents two separate considerations for users: model capability and the cost of accessing it. Muse Wonder Full’s results suggest Meta has produced a model that competes closely with leading systems on this evaluator’s tests. But its restricted availability means customers cannot purchase it through a per-token API, while the model that is available that way scored four points lower.
Artificial Analysis estimated that one Intelligence Index task costs $4.52 with Muse Wonder, compared with $0.72 for GPT-6.1 Sol at the same index score of 52. The evaluator also estimated $1.99 for Gemini 4 Argon, which scored 53, and $5.46 for Claude Sonnet 5.5, which scored 56. These are costs for the evaluator’s benchmark task, not a universal price for every customer workflow; actual bills depend on usage and token volumes.
The evaluation also reports limits relevant to deployment. Muse Wonder’s hallucination rate was 27%, while Wonder Full’s was 23%; the latter is an improvement from the 33% reported for Muse Spark 1.3, but neither figure means the models are reliable in every setting. Workflows where incorrect answers carry a cost may still need independent checks.
As an affiliate, we earn on qualifying purchases.
How the Two Models Differ
Meta’s launch follows Muse Spark 1.3 by six weeks, according to the source report. Artificial Analysis says Wonder Full improved on all ten evaluations in its Intelligence Index v4.3, while Wonder also recorded gains over Spark on several reported measures. For example, the API model scored 52% on Terminal-Bench and 68% on AutomationBench-AA, compared with 33% and 58%, respectively, for Spark 1.3.
The Intelligence Index combines results from multiple evaluations, so its overall score does not describe performance on every task. The source reports Muse Wonder Full at 30 and Muse Wonder at 25 on AA-Omniscience, a measure that rewards correct answers and penalizes confident errors. It lists higher scores for Claude Opus 5.5, GPT-6 Astra and Gemini 4 Argon. Results therefore vary by task, and an overall index position should not be read as a guarantee of superiority across uses.
Pricing comparisons also reflect how much text a model processes. Artificial Analysis says Muse Wonder used about 7.69 million input tokens per Intelligence Index task, compared with 6.03 million for Spark 1.3; it reports 54,000 output tokens for Wonder. The supplied source says input accounts for $3.43, or 76%, of the estimated $4.52 task cost. These figures help explain the evaluator’s estimate, but they may not represent typical customer workloads.
“The difference is the bill: one Intelligence Index task costs $4.52 on Muse Wonder and $0.72 on Sol.”
— Thorsten Meyer, ThorstenMeyerAI.com
As an affiliate, we earn on qualifying purchases.
Limits of the Published Comparison
The source material does not give a per-token price for Muse Wonder Full, which is available only through Muse Code. Its cost per task cannot be established from the information provided, and the Coding Agents Index score does not show how the model would perform in a different coding-agent setup.
The reported $4.52 figure is an evaluator’s estimate for one Intelligence Index task, not a fixed charge for all API requests. The source does not specify how closely that benchmark’s token use matches ordinary customer workloads. It also does not provide uncertainty ranges for every result in the summary, so small score differences should not automatically be treated as definitive rankings.
Meta’s material in the supplied source does not include a company statement explaining its pricing or the access limits on Wonder Full. The evaluation’s benchmark results and cost estimates are attributed to Artificial Analysis; broader performance in real-world use remains unestablished by these figures alone.
As an affiliate, we earn on qualifying purchases.
Access, Pricing and Real-World Use
Customers can assess Muse Wonder through Meta’s API and Muse Code at the listed token prices. Muse Wonder Full remains limited to Muse Code, according to the source. The supplied material does not give a date for wider availability, an API release or a pricing change.
The next useful evidence will come from more information about Wonder Full’s access and cost, alongside independent testing outside the Intelligence Index. For teams considering Muse Wonder, actual input and output volumes will matter: the evaluator’s estimate indicates that input processing made up most of the benchmark task bill. Users will also need to test models against their own tasks and retain checks where errors could have material consequences.
As an affiliate, we earn on qualifying purchases.
Key Questions
What are Muse Wonder and Muse Wonder Full?
They are two proprietary Meta models. Artificial Analysis scored Muse Wonder at 52 and Muse Wonder Full at 56 on its Intelligence Index v4.3.
Which model is available through Meta’s API?
Muse Wonder is available through Meta’s API and inside Muse Code. Muse Wonder Full is available only to Muse Code users, according to the supplied source.
How much does Muse Wonder cost?
Meta’s listed rates are $4 per million input tokens, $0.20 per million cached input tokens and $20 per million output tokens. Artificial Analysis estimated a cost of $4.52 for one of its Intelligence Index tasks; that is a benchmark-specific estimate, not a standard charge for every request.
Why can’t Muse Wonder Full’s API cost be compared?
The source says Wonder Full has no per-token price and is available only in Muse Code. It provides no equivalent API pricing or task-cost estimate.
Do the benchmark results establish that the models are reliable for every use?
No. The scores cover specific evaluations, and the source reports hallucination rates of 23% for Wonder Full and 27% for Wonder. Performance can vary by task, so users should test their own workflows and check outputs where errors matter.
Source: ThorstenMeyerAI.com
Columbus Day / Indigenous Peoples' Day Picks
long weekend sales
As an affiliate, we earn on qualifying purchases.
