🔍 Read the full analysis: Muse Wonder Full Joins The AI Frontier: Muse Wonder’s Charging Side on ThorstenMeyerAI.com
Get hardware and tech essentials 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 evaluated by Artificial Analysis on October 12, 2026. The evaluation gave Wonder Full a score of 56 and Wonder 52, but the API model’s estimated Intelligence Index task cost was $4.52, compared with $0.72 for GPT-6.1 Sol at the same score.
Meta has released Muse Wonder and Muse Wonder Full, and an Artificial Analysis evaluation published October 12, 2026 places the Full version among the top-scoring models tested. The API-available Muse Wonder scored 52 on the Intelligence Index, while the higher-scoring Full version scored 56 but is available only inside Meta’s Muse Code coding agent.
Artificial Analysis scored Muse Wonder Full at 56 on its Intelligence Index v4.3, eight points above Muse Spark 1.3’s 48. The publisher said Full improved on all ten evaluations in the index and ranked in the top three on six. Claude Opus 5.5 scored 58, while Claude Sonnet 5.5 also scored 56. These are the evaluation publisher’s results, not independent confirmation of performance in every real-world task.
Muse Wonder scored 52, matching GPT-6.1 Sol’s reported score. Meta makes Wonder available through its API and Muse Code; Wonder Full is available only to Muse Code users. Both models are proprietary, accept text, images, speech or audio, and video as input, return text, and have a one-million-token context window, according to the source report.
The reported API list price for Muse Wonder is $4 per million input tokens and $20 per million output tokens, with cached input listed at $0.20 per million tokens. Artificial Analysis estimated a cost of $4.52 for one Intelligence Index task with Wonder. The same report put GPT-6.1 Sol at $0.72 for a task and Claude Sonnet 5.5 at $5.46. These are publisher calculations based on its benchmark workload, not a guarantee of costs for every customer’s use.
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 gives developers and businesses a new model to evaluate, but the distinction between the two versions matters: the top-scoring model is not available through Meta’s API, while the API option posts a lower score and a comparatively high estimated benchmark cost. Buyers weighing models on price and performance should not assume that the Full version’s results describe Wonder’s API performance.
Artificial Analysis reported that Muse Wonder Full scored 57% on Terminal-Bench 4.0, up from 33% for Muse Spark 1.3, and that Muse Code paired with Wonder Full scored 68 in the Coding Agents Index. That tied the reported score for Claude Code paired with Claude Sonnet 5.5. The coding index assesses the model and its software harness together, so it does not establish how Wonder Full would perform in another coding environment.
The cost comparison also depends on the benchmark’s token usage. The report says a Muse Wonder task consumed 7.69 million input tokens and about 54,000 output tokens, with input accounting for $3.43, or 76%, of the $4.52 estimate. That makes the reported bill specific to the test workload; customer costs will vary with prompts, usage and cached inputs. The evaluation is useful for comparison, but it does not establish that one model is the better choice for every application.
As an affiliate, we earn on qualifying purchases.
How the Two Muse Models Differ
The release follows Muse Spark 1.3 by six weeks, according to the source report. Artificial Analysis gave Spark 1.3 an Intelligence Index score of 48 and estimated a $1.60 cost per index task. The publisher reported that Muse Wonder improved on Spark in several measures, including Terminal-Bench, where Wonder scored 52% compared with Spark’s 33%, and AutomationBench-AA, where Wonder scored 68% compared with 58%.
Wonder Full and Wonder share the reported input formats and context-window size, but differ in access and benchmark results. Artificial Analysis reported an AA-Omniscience score of 30 for Full and 25 for Wonder, and hallucination rates of 23% and 27%, respectively. The report said Spark’s hallucination rate was 33%. These rates are tied to the publisher’s evaluation and should not be treated as a prediction of error frequency in every use case.
Meta’s listed Wonder API rates are also higher than the prices cited in the report for several competitors. The source gives $2 per million input and $10 per million output tokens for GPT-6.1 Sol, Gemini 4 Argon and Claude Sonnet 5.5, while Wonder is listed at $4 and $20. A model’s actual cost depends on how much and what type of input and output a task requires, so list prices alone do not determine total operating cost.
As an affiliate, we earn on qualifying purchases.
What the Evaluation Cannot Show
The supplied report does not include a statement from Meta explaining the pricing, the limited availability of Wonder Full, or whether broader access is planned. It also provides no per-token price for Wonder Full, which is available only through Muse Code; its cost per task cannot be calculated from the information provided.
Benchmark results do not settle how the models will perform across customers’ workloads, and the report does not provide enough detail here to independently reproduce its estimated task costs. The cost comparison relies on the benchmark’s measured token consumption and stated list prices. It remains unclear whether Meta will change prices or access arrangements, and how performance will compare in routine use outside the evaluations.
The source material also ends during its account of the task-cost breakdown after specifying the input portion. It does not supply a complete breakdown of the remaining charges. No additional cost components or totals beyond the stated $4.52 estimate should be inferred from that incomplete detail.
As an affiliate, we earn on qualifying purchases.
Access, Pricing and Real-World Use
For now, the practical distinction is the access route: Muse Wonder is available through Meta’s API and inside Muse Code, while Wonder Full is limited to Muse Code users, according to the report. Developers considering Wonder can compare the published list rates with their expected token use and test the model against their own tasks before drawing conclusions from benchmark costs.
Further details that could clarify the release include whether Meta plans to offer Wonder Full through an API, how it would price that version, and whether Wonder’s rates will change. The supplied source does not announce a date for any such update. Users applying the models to consequential work should also account for the evaluation’s reported hallucination rates and maintain appropriate verification rather than relying on benchmark scores alone.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did Artificial Analysis score the new Muse models?
Artificial Analysis reported an Intelligence Index v4.3 score of 56 for Muse Wonder Full and 52 for Muse Wonder. The report says Full improved on all ten evaluations in the index.
Can developers access Muse Wonder Full through Meta’s API?
Not according to the supplied report. Muse Wonder Full is available only inside Muse Code; the standard Muse Wonder model is available through Meta’s API and Muse Code.
How much does Muse Wonder cost to use?
The reported list price is $4 per million input tokens, $0.20 per million cached input tokens, and $20 per million output tokens. Artificial Analysis estimated $4.52 for one task in its Intelligence Index benchmark; actual costs depend on usage.
Does Muse Wonder Full’s benchmark score prove it is better for every coding task?
No. The results describe specific evaluations. The Coding Agents Index measures a model-and-harness combination, and performance on a customer’s own tasks may differ.
Source: ThorstenMeyerAI.com
Columbus Day / Indigenous Peoples' Day Picks
long weekend sales
As an affiliate, we earn on qualifying purchases.
