🔍 Read the full analysis: Muse Wonder Full And Muse Wonder Join AI’s Charged-Up Frontier 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. Wonder Full scored 56 on the Intelligence Index, while API-available Wonder scored 52; the evaluation also found Wonder costs $4.52 per index task, 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 received substantially different access and pricing terms. In an evaluation published October 12 by Artificial Analysis, Wonder Full scored 56 on the Intelligence Index v4.3, while the API model Wonder scored 52; the evaluation estimated $4.52 per index task for Wonder, compared with $0.72 for GPT-6.1 Sol, which earned the same score.
Artificial Analysis reported that Muse Wonder Full scored 56, eight points above Muse Spark 1.3’s 48, and improved on all ten evaluations in the index. Among 27 configurations compared in the source report, Claude Opus 5.5 at its maximum setting scored 58, while Claude Sonnet 5.5 at its maximum setting also scored 56. Those results describe the evaluated configurations; they do not establish that the models perform identically across all uses.
The two Meta models have a one-million-token context window and accept text, images, speech or audio, and video as input, returning text. Access differs: Muse Wonder is available through Meta’s API and Muse Code, while Wonder Full is available only to Muse Code users. The report gives Wonder’s list prices as $4 per million input tokens, $0.20 per million cached input tokens, and $20 per million output tokens. Wonder Full has no per-token price listed, leaving its per-task cost unavailable to outside users.
Artificial Analysis’ individual benchmark results show Wonder Full placing in the top three on six of ten evaluations. Its Terminal-Bench 4.0 score was 57%, compared with 33% for Spark 1.3; it scored 1,857 Elo on GDPval-AA, close to the reported 1,866 for Opus 5.5 and 1,838 for Sonnet 5.5. The report also gave Wonder Full a 23% hallucination rate, down from 33% for Spark 1.3. Wonder scored 52% on Terminal-Bench and had a 27% hallucination rate.
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 Comes With a Price
The release gives developers a new option from Meta at a time when model quality, access and operating costs all affect product choices. The central distinction is between Wonder Full’s benchmark results and Wonder’s commercial availability: the stronger-scoring model is limited to Muse Code, while the API model is easier to integrate but has lower evaluation scores and a higher estimated task cost than several comparably scored rivals.
Artificial Analysis estimated that one Intelligence Index task costs $4.52 on Muse Wonder. Its comparison lists $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 Sonnet 5.5 at 56. These are benchmark-task estimates, not a universal price for every customer workload; actual bills depend on usage and token consumption. Still, they suggest that developers weighing the API model will need to compare both quality and costs on their own tasks.
The evaluation also points to progress in agent and coding work. In Artificial Analysis’ Coding Agents Index, Muse Code with Wonder Full scored 68, tied with Claude Code using Sonnet 5.5. That index evaluates the model together with its software harness. It does not show that Wonder Full would receive the same score in a different coding environment, nor that it is superior to the competing model on every task.
As an affiliate, we earn on qualifying purchases.
Two Models, Different Access
The launch follows Muse Spark 1.3 by six weeks, according to the source report. The report compares the new models with other evaluated systems using Artificial Analysis’ Intelligence Index v4.3, as well as separate measures of coding agents, knowledge work and hallucinations. Those measures offer snapshots of performance under defined tests, not guarantees about everyday outcomes.
The price comparison reflects both list rates and the amount of text models process in the index task. Artificial Analysis’ breakdown says Wonder used 7.69 million input tokens per task and about 54,000 output tokens. The report attributes $3.43, or 76% of the estimated $4.52 task cost, to input. For comparison, it says Spark 1.3 used 6.03 million input tokens and Wonder Full used 6.44 million. Token use and resulting costs may vary with task, prompts and deployment.
On knowledge reliability, the report’s AA-Omniscience scores were 30 for Wonder Full and 25 for Wonder. It lists higher scores for Opus 5.5 at 46, GPT-6 Astra at 43 and Gemini 4 Argon at 42. The report also compares hallucination rates among models scoring at least 45 on the Intelligence Index: 15% for Gemini 4 Argon, 23% for Wonder Full and 27% for Wonder. These figures apply to the evaluation method, not necessarily to all real-world interactions.
“The difference is the bill: one Intelligence Index task costs $4.52 on Muse Wonder and $0.72 on Sol.”
— Thorsten Meyer, ThorstenMeyerAI.com
high performance AI models for text and image processing
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Costs and Real-World Results
Wonder Full’s per-token or per-task price is not disclosed in the source material because it is available only through Muse Code. That prevents outside buyers from making a direct cost comparison with API models. The reported $4.52 for Wonder is an estimate for one Intelligence Index task, and the source does not establish that the same amount would apply to other workloads or usage patterns.
The evaluation does not settle how the models perform across every application, nor does a benchmark score establish reliability in a particular deployment. Hallucination rates and task scores depend on the test methods described by Artificial Analysis. The source report supplies no detailed methodology or confidence intervals for every result in the provided material, and it does not provide independent confirmation of Meta’s model specifications beyond the evaluation and product details it cites.
As an affiliate, we earn on qualifying purchases.
API Use and Further Evaluation
Developers can try Muse Wonder through Meta’s API or use either model through Muse Code where available, according to the source report. The next useful comparison will be testing the API model against alternatives on representative workloads, measuring both output quality and actual token bills rather than relying only on index-task estimates.
Further evaluation may clarify how Wonder Full performs outside Muse Code and whether Meta will make it available through an API or publish a price. The source material gives no timetable for broader access or additional releases. For now, the published evidence supports a performance comparison, while access and cost remain distinct limitations.
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 six weeks after Muse Spark 1.3, according to the source report. Both have a one-million-token context window, accept text, images, audio and video, and return text.
How did the models score?
Artificial Analysis reported scores of 56 for Muse Wonder Full and 52 for Muse Wonder on Intelligence Index v4.3. Wonder Full improved on all ten evaluations in the index, according to the report.
Which model can developers access through an 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 material.
Why is Muse Wonder’s estimated task cost drawing attention?
Artificial Analysis estimated a cost of $4.52 for one Intelligence Index task on Wonder. Its comparison puts GPT-6.1 Sol at $0.72 for the same score of 52, though costs for other workloads may differ.
What remains unknown about Muse Wonder Full?
The source does not give Wonder Full a per-token price or a per-task cost, and it provides no schedule for API availability. Its reported coding score also measures Muse Code and the model together, rather than the model in every possible coding environment.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
