MetaReasoning modelNewReleased 12 Oct 2026Compare with another model

Muse Wonder Full

Muse Wonder Full is a Meta proprietary reasoning model, released 12 Oct 2026. It is not yet ranked: 4 independent results so far, and the index needs at least five across two categories. Last measured 12 Oct 2026.

Blended price: –
Output speed: –
First answer: –
Context: 1M
Full answer: –
Cost per run: –
Released: 12 Oct 2026

How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.

Reasoning95
Knowledge75
Long context67
Composite92

Benchmark results

See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

Benchmark	max	Source
Humanity's Last Exam (AA)not in index	53.6%#16	Artificial Analysis
CritPt	32.9%#1	Artificial Analysis

Coding

Benchmark	max	Source
SciCode (AA)not in index	62.6%#4	Artificial Analysis
Terminal-Bench 4.0 (AA)not in index	57.1%#6	Artificial Analysis

Agents & tools

Benchmark	max	Source
GDPval-AA v2.1not in index	67.8%#2	Artificial Analysis
AutomationBenchnot in index	71.3%#3	Artificial Analysis
GDP.pdfnot in index	32.6%#1	Artificial Analysis
AA-Briefcase v1.1not in index	1729#5	Artificial Analysis

Knowledge

Benchmark	max	Source
AA-Omniscience	30.1#32	Artificial Analysis
AA-Omniscience: accuracynot in index	43.2%#98	Artificial Analysis
AA-Omniscience: non-hallucinationnot in index	77.0%#17	Artificial Analysis

Long context

Benchmark	max	Source
AA-LCR	83.7%#15	Artificial Analysis

Composite

Benchmark	max	Source
AA Intelligence Index v4.3.2	55.8#4	Artificial Analysis

Where it wins

Benchmarks where this configuration ranks in the top five of every configuration measured.