🔍 Read the full analysis: Muse Wonder Full Joins AI’s Frontier With Muse Wonder In Tow on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Meta has released Muse Wonder and Muse Wonder Full, two proprietary models rated by Artificial Analysis. Wonder Full scored 56 on the Intelligence Index, while the API-available Wonder scored 52; the evaluator estimated a $4.52 cost per index task for Wonder, far above some similarly rated competitors. Wonder Full is available only through Meta’s Muse Code, and its per-task cost is not publicly calculable from token pricing.
Meta has released Muse Wonder and Muse Wonder Full, two proprietary AI models that scored 52 and 56, respectively, on Artificial Analysis’ Intelligence Index v4.3. The independent evaluator’s October 12 assessment found broad gains over Muse Spark 1.3, while estimating that one index task with the API-available Muse Wonder costs $4.52—more than several competitors with similar or higher scores.
Artificial Analysis says Muse Wonder Full improved on all 10 evaluations in the index compared with Muse Spark 1.3, which scored 48. Wonder Full placed in the top three on six evaluations. Its Terminal-Bench 4.0 result rose from 33% to 57%, and its GDPval-AA v2.1 score was 1,857 Elo, near Claude Opus 5.5 at 1,866 and Claude Sonnet 5.5 at 1,838. The evaluator said those GDPval results fall within the models’ confidence intervals.
Availability differs between the two models. Muse Wonder is offered through Meta’s API and in Muse Code, while Muse Wonder Full is available only to Muse Code users. Both have a reported one-million-token context window, accept text, images, audio and video as input, and return text. Meta’s listed API rates for Wonder are $4 per million input tokens and $20 per million output tokens, with cached input priced at $0.20 per million tokens. Wonder Full has no per-token price, so its cost per task cannot be compared on the same basis from the information provided.
Artificial Analysis’ Coding Agents Index gave Muse Code with Wonder Full a score of 68, tied with Claude Code using Claude Sonnet 5.5. Muse Code with the API model Wonder scored 63, while the earlier Spark 1.3 configuration scored 54. These are evaluations of a model paired with its coding-agent software, not direct comparisons of the models operating in identical software environments.
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, Higher API Costs
The release gives developers evidence that Meta has made measurable progress in coding and knowledge-work evaluations. But for customers choosing an API, benchmark performance is only one part of the decision: usage cost and token consumption can affect whether a model is practical at scale.
Artificial Analysis estimated that one Intelligence Index task costs $4.52 with Muse Wonder. In the same comparison, GPT-6.1 Sol scored 52 at an estimated $0.72 per task; Gemini 4 Argon scored 53 at $1.99; GPT-6 Astra scored 53 at $3.26; and Claude Sonnet 5.5 scored 56 at $5.46. These figures are the evaluator’s estimates for its benchmark task, not a universal charge for every user prompt. They indicate how the models performed and were billed under that test’s conditions.
The evaluator attributed much of Wonder’s estimated bill to input processing: it used 7.69 million input tokens per index task, compared with 6.03 million for Spark 1.3. Artificial Analysis put input at $3.43, or 76% of Wonder’s estimated $4.52 task cost. The figures make token use a relevant consideration alongside Meta’s list prices. Actual costs will vary with workload, prompts, caching and usage patterns.
As an affiliate, we earn on qualifying purchases.
How The Models Compare
Meta released the models about six weeks after Muse Spark 1.3. In Artificial Analysis’ comparison of 27 model configurations, only Claude Opus 5.5 in its max configuration scored above Wonder Full’s 56, with a score of 58; Claude Sonnet 5.5 max also scored 56. Muse Wonder’s 52 matched GPT-6.1 Sol max in the index, according to the source material.
The benchmark gains were not limited to coding. Artificial Analysis reported that Wonder Full scored 1,729 on AA-Briefcase, third in the cited results, and led on GDP.pdf at 33% and CritPt at 33%. The evaluator also reported an AA-Omniscience score of 30 for Wonder Full and 25 for Wonder. It said Wonder Full’s hallucination rate was 23%, down from 33% for Spark 1.3; Wonder’s reported rate was 27%. These results reflect the evaluator’s particular tests and should not be treated as guarantees of accuracy in real-world use.
The models’ pricing also differs from the listed rates for several rivals cited in the report. Muse Spark 1.3 was listed at $1.25 per million input tokens and $4.25 per million output tokens; Muse Wonder is listed at $4 and $20. The source compares those prices with $2 per million input tokens and $10 per million output tokens for Claude Sonnet 5.5, Gemini 4 Argon and GPT-6.1 Sol. The benchmark task-cost estimates incorporate usage in the evaluator’s test, rather than reflecting token prices alone.
AI input output text video audio tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Costs And Real-World Accuracy
The evaluation does not establish how Wonder and Wonder Full will perform across every customer workload, or whether the benchmark cost estimates will match users’ bills. The source does not provide a per-task price for Wonder Full, which is restricted to Muse Code, so direct cost comparisons for that model remain unavailable.
It is also unclear from the supplied material how Meta’s models were trained, what safeguards or deployment limits apply, and whether the company plans to broaden Wonder Full’s access. The reported hallucination rates are benchmark measurements, not assurances that a model will avoid errors in other settings. Artificial Analysis’ comparisons and scores should be read as results from its stated index and test configurations, rather than a complete account of product performance.
As an affiliate, we earn on qualifying purchases.
Access And Pricing To Watch
Developers can access Muse Wonder through Meta’s API and Muse Code, while Muse Wonder Full is currently limited to Muse Code users. The next developments to watch are whether Meta expands access to the higher-scoring model, publishes a pricing method that allows customers to estimate its cost, or updates Wonder’s token rates.
For now, potential API users can compare Meta’s published per-token prices with their expected input and output volumes, and test the model on their own tasks before committing to larger workloads. Further independent evaluations and user experience data will help show whether the benchmark gains translate into reliable performance and acceptable costs outside the test environment.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did Meta release?
Meta released Muse Wonder and Muse Wonder Full, proprietary models with a reported one-million-token context window and support for text, images, audio and video inputs.
Which model can developers access through Meta’s API?
Muse Wonder is available through Meta’s API and in Muse Code. Muse Wonder Full is currently available only to Muse Code users.
How did the models score?
Artificial Analysis gave Muse Wonder Full a score of 56 and Muse Wonder a score of 52 on its Intelligence Index v4.3. Muse Spark 1.3 scored 48 in the cited comparison.
How much does Muse Wonder cost?
Meta lists Muse Wonder at $4 per million input tokens and $20 per million output tokens, with cached input at $0.20 per million tokens. Artificial Analysis estimated a $4.52 cost for one of its index tasks; actual API costs vary by usage.
Is Muse Wonder Full’s cost known?
The supplied information gives no per-token price for Muse Wonder Full. Because the model is limited to Muse Code, its cost per task cannot be calculated from the API rates listed for Muse Wonder.
Source: ThorstenMeyerAI.com
Columbus Day / Indigenous Peoples' Day Picks
long weekend sales
As an affiliate, we earn on qualifying purchases.
