🔍 Read the full analysis: Muse Wonder Full Joins AI’s Frontier: The Muse Wonder Charging Angle 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
Artificial Analysis’ October 12 evaluation gave Meta’s Muse Wonder Full an Intelligence Index score of 56, while the API-available Muse Wonder scored 52. The results put Meta’s coding system among frontier-level configurations, but the evaluation also found Muse Wonder costly per task compared with models at similar scores.
Meta’s Muse Wonder Full scored 56 on Artificial Analysis’ Intelligence Index v4.3, in an evaluation published October 12, while the API model Muse Wonder scored 52. The results place Meta’s new models near leading systems on the benchmark, but the evaluation estimates that one Muse Wonder task costs $4.52—far more than a similarly scoring GPT-6.1 Sol task.
Artificial Analysis reported that Muse Wonder Full improved on all 10 evaluations in its index compared with Muse Spark 1.3. It ranked in the top three on six evaluations. Its Terminal-Bench 4.0 score rose to 57%, from 33% for Muse Spark 1.3, while its GDPval-AA v2.1 result was 1,857 Elo, close to scores reported for Claude Opus 5.5 and Claude Sonnet 5.5. The source article says those GDPval-AA results fall within the models’ confidence intervals.
The two models are proprietary and support a one-million-token context window, with text, image, speech or audio, and video inputs; both return text. Muse Wonder is available through Meta’s API and in Muse Code. Meta’s listed API prices are $4 per million input tokens, $0.20 per million cached input tokens and $20 per million output tokens. Muse Wonder Full is available only to Muse Code users and has no public per-token price, so its cost per task cannot be calculated from the information provided.
On the Coding Agents Index, which evaluates the model together with its software harness, Muse Code plus Wonder Full scored 68, matching Claude Code plus Sonnet 5.5. Muse Code plus the API-available Wonder scored 63. These are combined-system results, not evidence that Wonder Full would achieve the same score in another coding environment.
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 High API Costs
The results suggest Meta has made a substantial benchmark advance in a short time: Muse Wonder Full’s overall score is eight points above Muse Spark 1.3’s 48, and its terminal-task score rose by 24 percentage points. For developers, the split between the two models matters: the stronger system is limited to Muse Code, while the model available through the API has a lower index score and a measurable usage price.
Artificial Analysis estimates $4.52 per Intelligence Index task for Muse Wonder, compared with $0.72 for GPT-6.1 Sol at the same score of 52. The comparison is based on the evaluation’s task-cost estimates, not necessarily the price of every real-world workload. Still, it gives businesses a reason to weigh performance against token consumption and API pricing before choosing a model. The source says Muse Wonder uses 7.69 million input tokens per index task, with input accounting for $3.43, or 76%, of the estimated total cost.
Accuracy remains relevant alongside benchmark scores. Artificial Analysis placed Muse Wonder’s hallucination rate at 27% and Wonder Full’s at 23%, down from 33% for Muse Spark 1.3 according to the source. The article reports a 15% rate for Gemini 4 Argon among models scoring at least 45 on the index. These figures come from the cited evaluation and do not establish how a system will perform in every application; users handling consequential work may still need independent checks.
As an affiliate, we earn on qualifying purchases.
Two Models, Different Access
Meta released Muse Wonder and Muse Wonder Full about six weeks after Muse Spark 1.3, according to the October 13 source article. Their shared technical description includes multimodal input and a one-million-token context window, but access differs: Wonder is sold through the API and offered in Muse Code, whereas Wonder Full is limited to Muse Code. The lack of public per-token pricing for Wonder Full leaves its operating cost unclear.
The Artificial Analysis Intelligence Index compares models across ten evaluations. The source reports Wonder Full at 56, against 58 for Claude Opus 5.5 and 56 for Claude Sonnet 5.5; Muse Wonder’s 52 matches the reported score for GPT-6.1 Sol. The source also cites task-cost estimates of $5.98 for Opus 5.5 and $5.46 for Sonnet 5.5. These numbers describe the index’s tested configurations and estimated task costs, rather than a universal price for using each model.
Pricing changes from Meta’s prior model are also part of the comparison. The source lists Muse Spark 1.3 at $1.25 per million input tokens and $4.25 per million output tokens, versus Wonder’s $4 and $20 rates. It reports that other named models, including Sonnet 5.5, Gemini 4 Argon and GPT-6.1 Sol, list at $2 per million input tokens and $10 per million output tokens. Actual bills depend on token use, and the source’s task-cost analysis indicates that input consumption is a major factor for Wonder.
“The gap is the story.”
— Thorsten Meyer, ThorstenMeyerAI.com
As an affiliate, we earn on qualifying purchases.
Costs and Real-World Results
The evaluation provides no public per-token price for Muse Wonder Full, preventing a direct cost-per-task comparison with API models. The available material also does not give enough detail to independently reproduce the reported benchmark scores or estimated task costs, including the full cost breakdown beyond the stated input share for Muse Wonder.
Benchmark rankings do not establish performance across every use case. The source reports that Wonder Full’s Coding Agents Index result combines Muse Code with the model; it does not show how the model would perform with a different harness. The evaluation’s hallucination rates and task costs likewise reflect the test setup, and the material does not specify how closely those conditions match particular customers’ workloads.
It is also unclear whether Meta will make Wonder Full available through its API, publish its pricing, or change Wonder’s list prices. The source’s comparisons are a snapshot of the models and rates reported in October 2026; future versions, updated evaluations or pricing changes could alter them.
As an affiliate, we earn on qualifying purchases.
Watch for Access and Pricing Changes
The next practical developments to watch are whether Meta expands Wonder Full beyond Muse Code and whether it publishes a price that permits direct cost comparisons. For API users, the immediate choice is between the reported Wonder capabilities and its list price, with the evaluation suggesting that input-token use contributes most to its benchmark task cost.
Further independent testing could show whether the reported gains carry over to other workloads and coding environments. Until then, readers should treat the Intelligence Index and Coding Agents Index as results for the specific evaluations described, rather than guarantees of performance or cost in production. Meta’s next model release or pricing announcement was not specified in the source material.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did Artificial Analysis report about Muse Wonder Full?
It reported an Intelligence Index v4.3 score of 56 and said the model improved on all 10 index evaluations compared with Muse Spark 1.3. The evaluation was published October 12, 2026.
Can developers access Muse Wonder Full through Meta’s API?
Not according to the source article. It says Muse Wonder Full is available only to Muse Code users; Muse Wonder is the version available through Meta’s API.
How much does Muse Wonder cost?
Meta’s listed rates in the source are $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 Intelligence Index task, a test-specific figure rather than a fixed charge for every use.
Does Muse Wonder Full tie Claude Opus 5.5?
Not on the overall Intelligence Index score cited: Wonder Full scored 56, compared with 58 for Claude Opus 5.5. In the Coding Agents Index, Muse Code plus Wonder Full scored 68, matching the reported score for Claude Code plus Claude Sonnet 5.5. That index measures model-and-harness combinations.
What remains unknown about the release?
The source does not provide a public per-token price for Wonder Full or say whether Meta plans to release it through the API. The extent to which the benchmark scores and estimated costs reflect specific real-world workloads is also not established.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
