Muse Wonder Full Ranks Fifth on Vals Index While Matching Claude Opus 5.5's List Price
The November 5 update puts Wonder Full behind four rivals on the composite index, even as it keeps its lead on sustained coding work
By Icarus3 min read
Independent benchmark results published this week place Meta’s Muse Wonder Full fifth on Vals AI’s GDP-weighted Vals Index, behind Claude Sonnet 5.5, Gemini 4 Carbon, Claude Opus 5.5, and Gemini 4 Argon. The ranking reorders the table Meta presented at Wonder’s October 12 launch, where its own benchmarks put Wonder Full ahead of Opus 5.5. On price, the two models are identical: Muse Wonder’s API list price is $4 per million input tokens and $20 per million output tokens, the same rates Anthropic lists for Opus 5.5.
What the Vals Results Show
The newest numbers come from Finance Agent v2.1, Vals AI’s revision of its financial-analyst benchmark, dated November 5. Claude Opus 5.5 leads at 67.85%, followed by Gemini 4 Argon at 67.66% and Muse Wonder Full at 67.58%. The standard Muse Wonder sits eighth at 65.5%. The gap from first to third is 0.27 points. Vals AI says v2.1 scores are not comparable with the earlier v2, which used different questions and rubrics.
On the earlier Finance Agent v2, dated October 29, the spread was wider: Gemini 4 Argon led at 65.4%, with Wonder Full sixth at 59.1% and the standard Wonder at 56.6%. Finance Agent tests the work of a junior financial analyst, pulling figures from filings, computing ratios, and reconciling numbers, with tools available.
The composite view is the Vals Index, which weights finance, coding, legal, and tax benchmarks by each sector’s share of US GDP. In the November 5 update, Claude Sonnet 5.5 leads at 70.8%, ahead of Gemini 4 Carbon at 70.5%, Claude Opus 5.5 at 70.2%, and Gemini 4 Argon at 69.2%. Muse Wonder Full places fifth at 66.8%. On the Vals Finance Index, a finance-only composite updated November 5, Wonder Full is fourth at 64.5%, behind Gemini 4 Carbon at 67.05%, Opus 5.5 at 65.59%, and Gemini 4 Argon at 64.89%.
How That Compares With Meta’s Launch Table
Meta’s launch chart, shared by Mark Zuckerberg on October 12, showed different numbers. On Meta’s internal Personal Superintelligence Index, Wonder Full scored 59.1 against 55.2 for Opus 5.5 and 54.5 for GPT-6 Astra. The legal-agent benchmark Harvey LAB-AA showed a wider gap, 13.1 for Wonder Full against 4.2 to 8.6 for rivals, and on MemoryBench both Wonder models scored around 85.5, ahead of Opus 5.5 at 81.5. On coding, Wonder Full led SWE-Marathon 57.5 to 46.3 and DeepSWE 1.1 79.0 to 74.2. Meta’s chart did show Opus 5.5 ahead on ProgramBench and HealthBench Pro, and GPT-6 Astra on top of the GDP.pdf professional-work test.
Independent numbers at launch were already closer. On the Artificial Analysis Intelligence Index, Wonder Full scored 56, behind Opus 5.5 at 58, with the standard Wonder at 52.
The Counterpoint: Wonder Still Leads on Sustained Coding
Coding runs the other way. Wonder Full leads Vals AI’s Vibe Code Bench for sustained coding work, scoring 33.98% against 30.36% for Opus 5.5, and Muse Wonder tied for second on DualEntry’s accounting benchmark. Wonder Full debuted sixth on LMArena’s WebDev leaderboard. Finance and professional-work benchmarks run against Meta’s launch table; coding still supports it.
The Price Picture
Muse Wonder’s API rates are $4 per million input tokens and $20 per million output tokens, as detailed in our pricing breakdown. Anthropic lists the same $4 and $20 rates for Opus 5.5, checked against its pricing page on November 3. Wonder Full is available only through Muse Code’s Power Usage ($50 a month) and Maximum Usage ($200 a month) plans, not through the API.
Google’s Gemini 4 Argon launched at an introductory $2 per million input tokens and $10 per million output tokens, half Wonder’s rates, through November 30. The price doubles to $4 and $20 on December 1, the same rates Wonder and Opus 5.5 already charge.
