🔍 Read the full analysis: Muse Wonder Full Joins The AI Frontier: A New Chapter For Muse Wonder on ThorstenMeyerAI.com
Get everyday 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, according to ThorstenMeyerAI.com, which reports that Artificial Analysis scored them at 52 and 56. The stronger model is limited to Muse Code, while the API model’s reported benchmark-task cost is $4.52—more than several competitors with similar or higher scores.
Meta has released Muse Wonder and Muse Wonder Full, two proprietary AI models that, according to an evaluation by Artificial Analysis reported by ThorstenMeyerAI.com, score 52 and 56 on the Intelligence Index. The release puts the stronger model in Meta’s coding agent rather than its public API, while the API model carries a reported benchmark-task cost of $4.52—a key consideration for developers comparing it with competing systems.
Artificial Analysis published its evaluation on October 12, 2026. The report says Muse Wonder Full scored 56 on the Intelligence Index v4.3, up from 48 for Muse Spark 1.3, and improved on all ten evaluations in the index. Among the 27 configurations compared by the report, Claude Opus 5.5 scored 58, while Claude Sonnet 5.5 also scored 56. Muse Wonder scored 52, matching the reported score for GPT-6.1 Sol.
The two models differ in access. Muse Wonder is available through Meta’s API and Muse Code, while Muse Wonder Full is available only to Muse Code users. The report lists Muse Wonder at $4 per million input tokens, $0.20 per million cached input tokens and $20 per million output tokens. Muse Wonder Full has no per-token price in the report, which says its cost per task cannot be calculated from public pricing.
On individual evaluations, the report gives Muse Wonder Full a 57% Terminal-Bench score, compared with 33% for Muse Spark 1.3, and places it among the top three on six of the ten tests. It also reports a 23% hallucination rate for Wonder Full and 27% for Wonder, versus 33% for Spark 1.3. The report cautions that the lower rates do not remove the need to check outputs in workflows where incorrect answers carry costs.
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
Performance Meets API Cost
The release’s practical importance depends on which model users can access and what they are willing to pay. The reported 56 score for Wonder Full suggests Meta has a model that performs near leading systems on this index, but its Muse Code-only availability limits direct use by developers seeking a priced API product. The API model, Muse Wonder, is more accessible but scores four points lower.
ThorstenMeyerAI.com reports that one Intelligence Index task costs $4.52 on Muse Wonder, compared with $0.72 for GPT-6.1 Sol at the same score of 52, $1.99 for Gemini 4 Argon at 53, and $5.46 for Claude Sonnet 5.5 at 56. These figures are the report’s benchmark-task estimates, not a universal charge for every user request. Actual spending will depend on workload and token use.
The gap matters for organizations evaluating models on both quality and operating expense. The evaluation suggests Muse Wonder is a stronger performer than its predecessor, but the reported price comparison raises a separate question: whether its performance and product features justify its cost for a particular task. The benchmark alone cannot answer that for every deployment.
As an affiliate, we earn on qualifying purchases.
From Spark to Two Models
The report describes the launch as coming six weeks after Muse Spark 1.3. Both new models are said to accept text, images, speech or audio, and video as input, return text, and support a one-million-token context window. The source identifies them as proprietary models.
Artificial Analysis also compared coding-agent configurations, which combine a model with its software harness. It reports 68 for Muse Code with Wonder Full, tying Claude Code with Sonnet 5.5; Claude Code with Opus 5.5 scored 66. Muse Code with Wonder scored 63, against 54 for Muse Code with Spark 1.3. The report stresses that these are model-and-harness results, not a direct test of Wonder Full inside Claude Code.
The reported cost estimate reflects more than list price. Muse Wonder used 7.69 million input tokens per index task, according to the source, compared with 6.03 million for Spark 1.3. The report attributes $3.43, or 76% of Wonder’s $4.52 task cost, to input tokens. This helps explain why token usage and pricing both shape the comparison.
“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.
Costs and Access Still Differ
The source provides no per-token price for Muse Wonder Full, so its cost cannot be compared on the same basis as the API model. It also does not establish when or whether Full might become available through the API. Meta’s own product announcement, pricing documentation and availability plans are not included in the supplied material.
The $4.52 figure is an estimate for one task in the Artificial Analysis Intelligence Index evaluation, as reported by Meyer. It should not be read as a fixed price for an ordinary prompt or as a guarantee of cost across different workloads. The source excerpt also ends during its token-cost breakdown, so it does not provide a complete accounting of every component beyond the stated input share.
Benchmark results describe performance on selected tests, not outcomes across all real-world uses. The supplied report gives scores and comparisons but does not include confidence intervals for every result or enough detail here to independently assess the evaluation methodology. The performance and cost findings should therefore be attributed to Artificial Analysis as relayed by the source.
As an affiliate, we earn on qualifying purchases.
Watch for Pricing and API Access
The next concrete developments to watch are whether Meta publishes more detail on Wonder Full pricing, expands its availability beyond Muse Code, or updates API rates for Muse Wonder. Those changes would affect how developers can compare the models on operating cost and access, beyond the published benchmark scores.
For now, developers considering the API model can compare the reported benchmark results with their own workloads, including token volume and the need to verify outputs. Further independent evaluations or results from real deployments could clarify whether Wonder’s higher reported task cost is offset by capabilities that matter to particular applications. No such pricing change, wider release or follow-up evaluation is confirmed in the supplied source.
As an affiliate, we earn on qualifying purchases.
Key Questions
What are Muse Wonder and Muse Wonder Full?
They are two proprietary Meta models reported to accept text, images, audio or speech, and video as input, with a one-million-token context window. Both return text, according to the source.
How did the models score?
Artificial Analysis scored Muse Wonder Full at 56 and Muse Wonder at 52 on Intelligence Index v4.3, as reported by ThorstenMeyerAI.com. The same report gives Claude Opus 5.5 a 58 and Claude Sonnet 5.5 a 56.
Can developers buy Muse Wonder Full through Meta’s API?
The supplied report says no: Wonder Full is available only to Muse Code users. Muse Wonder is the model listed as available through Meta’s API and Muse Code.
Why is Muse Wonder’s reported task cost a concern?
The report estimates one Intelligence Index task at $4.52 for Muse Wonder, compared with $0.72 for GPT-6.1 Sol, which scored 52 in the same comparison. These are evaluation-specific estimates, not guaranteed costs for all user requests.
What remains unknown about the release?
The supplied material does not give Wonder Full a per-token price or say whether it will reach Meta’s API. It also does not establish how the benchmark results will translate to every real-world workload.
Source: ThorstenMeyerAI.com
Columbus Day / Indigenous Peoples' Day Picks
long weekend sales
As an affiliate, we earn on qualifying purchases.
