Head-to-head comparison

Muse Wonder vs. Muse Wonder Full

Compare published benchmark results from matching versions and cohorts, API pricing, and technical specifications. Any documented test setup differences remain visible.

shared result series: 180
distinct benchmarks: 41
latest retrieval: 2026-10-15

First modelSecond model

Specifications

General specifications

Release date, availability, context window, architecture, and published parameter counts at a glance. Undisclosed details remain marked as unavailable.

Specifications for Muse Wonder and Muse Wonder Full
Attribute	Muse Wonder	Muse Wonder Full
Provider	Meta	Meta
Released	Oct 12, 2026	Oct 12, 2026
Availability	Preview	Subscription only
Model type	Proprietary	Proprietary
Parameters	Not published	Not published
Architecture	Not published	Not published
Context window	1,048,576 tokens	1M tokens
Knowledge cutoff	Not published	Not published
Technical sources	Model, Context	Model, Context

Pricing

API pricing compared

The table compares official standard API rates per 1M tokens, including published cache pricing. Batch, fast, priority, regional surcharges, tool calls, and cloud platform pricing are excluded.

API pricing for Muse Wonder and Muse Wonder Full
Price type	Muse Wonder	Muse Wonder Full
API input	$4	—
API output	$20	—
Cache read	$0.2	Not listed
Cache write (5 min.)	Not listed	Not listed
Cache write (1 hr.)	Not listed	Not listed
Price verified	10/15/2026 Source	10/15/2026

Pricing by provider

Direct developer pricing and provider endpoint tariffs routed through OpenRouter are listed separately. Prices are in USD per 1M tokens. Context-dependent direct tiers plus regional, Flex, Priority, and other OpenRouter tariffs remain individually identifiable.

No provider tariff for Muse Wonder is available in the stored OpenRouter snapshot. The official direct price is still shown.

No provider tariff for Muse Wonder Full is available in the stored OpenRouter snapshot. The official direct price is still shown.

Provider pricing for Muse Wonder and Muse Wonder Full
Model	Provider	Purchase route and tariff	Input per 1M tokens	Output per 1M tokens	Source
Muse Wonder	Meta	Direct from developerStandard API	Lowest listed price: $4	Lowest listed price: $20	Source10/15/2026

Performance

Shared benchmarks

Only published results with the same benchmark version, task, metric, and comparison cohort are paired. Different reasoning levels, agents, harnesses, or output limits appear directly in the table.

Capability profile from matched benchmarks

Each axis averages directly matched benchmark families. A value of 100 means the stronger value within this pair, not a universal quality score.

Muse Wonder

Muse Wonder Full

Coding: 90.9 / 99.8 (11 benchmark families)

Reasoning: 94.6 / 100 (3 benchmark families)

Mathematics: 89.4 / 100 (1 benchmark family)

Knowledge: 92.3 / 99.9 (2 benchmark families)

Finance: 93.4 / 99.9 (4 benchmark families)

Legal: 85.5 / 98.4 (3 benchmark families)

Agentic tasks: 92.1 / 100 (6 benchmark families)

Productivity: 93.9 / 100 (3 benchmark families)

180 matched rows from 41 benchmark families were considered. The radar shows 8 of 12 comparable categories.

Muse WonderMuse Wonder Full

AA-BriefcaseOverall. Higher is better. No documented setup difference

Muse Wonder1,658.65 elo

Muse Wonder Full1,729.32 elo

AA-LCROverall. Higher is better. No documented setup difference

Muse Wonder80.7%

Muse Wonder Full83.7%

AA-OmniscienceOverall. Higher is better. No documented setup difference

Muse Wonder25.483 points

Muse Wonder Full30.15 points

Artificial Analysis Intelligence IndexOverall. Higher is better. No documented setup difference

Muse Wonder52.394 points

Muse Wonder Full55.802 points

AutomationBench-AAOverall. Higher is better. No documented setup difference

Muse Wonder68.1%

Muse Wonder Full71.3%

AutomationBench-AA (strict pass@1)Overall. Higher is better. No documented setup difference

Muse Wonder39.1 %

Muse Wonder Full45.4 %

BioMysteryBenchOverall. Higher is better. No documented setup difference

Muse Wonder67.037 %

Muse Wonder Full73.333 %

Code MigrationOverall. Higher is better. No documented setup difference

Muse Wonder62.552 %

Muse Wonder Full61.033 %

CyberBench v1.1Overall. Higher is better. No documented setup difference

Muse Wonder80.357 %

Muse Wonder Full73.631 %

DeepSWEOverall. Higher is better. No documented setup difference

Muse Wonder78.8 %

Muse Wonder Full79 %

EMBOverall. Higher is better. No documented setup difference

Muse Wonder67.084 %

Muse Wonder Full71.917 %

Finance Agent (v2)Overall. Higher is better. No documented setup difference

Muse Wonder56.637 %

Muse Wonder Full59.061 %

Key to the marks:better valuebetter, but not conclusivetie

Benchmark scores for Muse Wonder and Muse Wonder Full
Benchmark and task	Muse Wonder	Muse Wonder Full	Source
AA-BriefcaseOverall Test details No documented difference in the test setupVersion: v1.1Metric: AA-Briefcase EloScoring: Higher is betterMuse Wonder: Snapshot: Muse Wonder (Max), Reasoning: Max Muse Wonder Full: Snapshot: Muse Wonder Full (Max), Reasoning: MaxMuse Wonder: Exact AA configuration: Muse Wonder (Max) (release muse-wonder, row muse-wonder, UUID ff2f3f3d-783f-4e27-8a39-47237501a313, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.Muse Wonder Full: Exact AA configuration: Muse Wonder Full (Max) (release muse-wonder-full, row muse-wonder-full, UUID e994a6e0-d36b-4dab-8ae1-c1b469e4d9e2, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.	1,658.65 elo	Winner1,729.32 elo	Artificial Analysis Source details Data as of: 10/15/2026
AA-LCROverall Test details No documented difference in the test setupVersion: v1.1Metric: Pass rateScoring: Higher is betterMuse Wonder: Snapshot: Muse Wonder (Max), Reasoning: Max Muse Wonder Full: Snapshot: Muse Wonder Full (Max), Reasoning: MaxMuse Wonder: Exact AA configuration: Muse Wonder (Max) (release muse-wonder, row muse-wonder, UUID ff2f3f3d-783f-4e27-8a39-47237501a313, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.Muse Wonder Full: Exact AA configuration: Muse Wonder Full (Max) (release muse-wonder-full, row muse-wonder-full, UUID e994a6e0-d36b-4dab-8ae1-c1b469e4d9e2, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.	80.7%	Winner83.7%	Artificial Analysis Source details Data as of: 10/15/2026
AA-OmniscienceOverall Test details No documented difference in the test setupVersion: unspecifiedMetric: AA-Omniscience IndexScoring: Higher is betterMuse Wonder: Snapshot: Muse Wonder (Max), Reasoning: Max Muse Wonder Full: Snapshot: Muse Wonder Full (Max), Reasoning: MaxMuse Wonder: Exact AA configuration: Muse Wonder (Max) (release muse-wonder, row muse-wonder, UUID ff2f3f3d-783f-4e27-8a39-47237501a313, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.Muse Wonder Full: Exact AA configuration: Muse Wonder Full (Max) (release muse-wonder-full, row muse-wonder-full, UUID e994a6e0-d36b-4dab-8ae1-c1b469e4d9e2, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.	25.483 points The index can be negative and ranges from -100 to 100 according to Artificial Analysis.	Winner30.15 points The index can be negative and ranges from -100 to 100 according to Artificial Analysis.	Artificial Analysis Source details Data as of: 10/15/2026
Artificial Analysis Intelligence IndexOverall Test details No documented difference in the test setupVersion: v4.3.2Metric: Artificial Analysis Intelligence IndexScoring: Higher is betterMuse Wonder: Snapshot: Muse Wonder (Max), Reasoning: Max Muse Wonder Full: Snapshot: Muse Wonder Full (Max), Reasoning: MaxMuse Wonder: Exact AA configuration: Muse Wonder (Max) (release muse-wonder, row muse-wonder, UUID ff2f3f3d-783f-4e27-8a39-47237501a313, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.Muse Wonder Full: Exact AA configuration: Muse Wonder Full (Max) (release muse-wonder-full, row muse-wonder-full, UUID e994a6e0-d36b-4dab-8ae1-c1b469e4d9e2, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.	52.394 points	Winner55.802 points	Artificial Analysis Source details Data as of: 10/15/2026
AutomationBench-AAOverall Test details No documented difference in the test setupVersion: unspecifiedMetric: Objectives completed without guardrail violationsScoring: Higher is betterMuse Wonder: Snapshot: Muse Wonder (Max), Reasoning: Max Muse Wonder Full: Snapshot: Muse Wonder Full (Max), Reasoning: MaxMuse Wonder: Exact AA configuration: Muse Wonder (Max) (release muse-wonder, row muse-wonder, UUID ff2f3f3d-783f-4e27-8a39-47237501a313, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.Muse Wonder Full: Exact AA configuration: Muse Wonder Full (Max) (release muse-wonder-full, row muse-wonder-full, UUID e994a6e0-d36b-4dab-8ae1-c1b469e4d9e2, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.	68.1%	Winner71.3%	Artificial Analysis Source details Data as of: 10/15/2026
AutomationBench-AA (strict pass@1)Overall Test details No documented difference in the test setupVersion: Meta Wonder scorecard; version unspecifiedMetric: strict pass@1; every end-state assertion passedScoring: Higher is betterSettings: Reasoning: maxBusiness workflows in a simulated application environment. Every end-state assertion must pass; max reasoning. Do not merge with other AutomationBench-AA metrics.	39.1 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Winner45.4 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Meta Source details Data as of: 10/15/2026
BioMysteryBenchOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Harness: Terminus 2, Temperature: 1, Output limit: 131,072 tokens	67.037 %	Winner73.333 %	Vals AI Source details Data as of: 10/12/2026
Code MigrationOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	Winner62.552 %	61.033 %	Vals AI Source details Data as of: 10/12/2026
CyberBench v1.1Overall Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	Winner80.357 %	73.631 %	Vals AI Source details Data as of: 10/12/2026
DeepSWEOverall Test details No documented difference in the test setupVersion: v1.1Metric: pass@1Scoring: Higher is betterSettings: Reasoning: max113 tasks across 91 repositories and five languages. Max reasoning. Meta attributes these results to Datacurve; the Muse scores could not be found on its public leaderboard at retrieval.	78.8 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Winner79 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Meta Source details Data as of: 10/15/2026
EMBOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	67.084 %	Winner71.917 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Overall Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	56.637 %	Winner59.061 %	Vals AI Source details Data as of: 10/12/2026
GDP.pdfOverall Test details No documented difference in the test setupVersion: unspecifiedMetric: All-pass rateScoring: Higher is betterMuse Wonder: Snapshot: Muse Wonder (Max), Reasoning: Max Muse Wonder Full: Snapshot: Muse Wonder Full (Max), Reasoning: MaxMuse Wonder: Exact AA configuration: Muse Wonder (Max) (release muse-wonder, row muse-wonder, UUID ff2f3f3d-783f-4e27-8a39-47237501a313, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.Muse Wonder Full: Exact AA configuration: Muse Wonder Full (Max) (release muse-wonder-full, row muse-wonder-full, UUID e994a6e0-d36b-4dab-8ae1-c1b469e4d9e2, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.	30.6%	Winner32.6%	Artificial Analysis Source details Data as of: 10/15/2026
GDP.pdfOverall Test details No documented difference in the test setupVersion: Meta Wonder scorecard; version unspecifiedMetric: published GDP.pdf scoreScoring: Higher is betterSettings: Reasoning: maxProfessional work grounded in supplied PDFs. Max reasoning. Meta names Surge as the original source.	30.8 points Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort. The source specifies neither a denominator nor a normalized scale. This is the published raw score, not a percentage.	Winner33.2 points Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort. The source specifies neither a denominator nor a normalized scale. This is the published raw score, not a percentage.	Meta Source details Data as of: 10/15/2026
GDPval-AAOverall Test details No documented difference in the test setupVersion: v2.1Metric: GDPval-AA EloScoring: Higher is betterMuse Wonder: Snapshot: Muse Wonder (Max), Reasoning: Max Muse Wonder Full: Snapshot: Muse Wonder Full (Max), Reasoning: MaxMuse Wonder: Exact AA configuration: Muse Wonder (Max) (release muse-wonder, row muse-wonder, UUID ff2f3f3d-783f-4e27-8a39-47237501a313, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.Muse Wonder Full: Exact AA configuration: Muse Wonder Full (Max) (release muse-wonder-full, row muse-wonder-full, UUID e994a6e0-d36b-4dab-8ae1-c1b469e4d9e2, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.	1,770.34 elo	Winner1,857.09 elo	Artificial Analysis Source details Data as of: 10/15/2026
Harvey LAB-AAOverall Test details No documented difference in the test setupVersion: 1.1Metric: all criteria passed with hallucination checksScoring: Higher is betterSettings: Reasoning: maxLegal deliverables must pass every criterion including hallucination checks. Max reasoning. Meta attributes the original evaluation to Artificial Analysis.	10 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Winner13.1 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Meta Source details Data as of: 10/15/2026
Harvey's Legal Agent BenchmarkOverall · Task fully resolved Test details No documented difference in the test setupVersion: 1Metric: task resolution rateScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	17.083 %	Winner21.667 %	Vals AI Source details Data as of: 10/12/2026
HealthBench ProOverall Test details No documented difference in the test setupVersion: Meta Wonder scorecard; version unspecifiedMetric: length-normalized weighted rubric scoreScoring: Higher is betterSettings: Reasoning: max525 health questions without tools; physician rubrics graded by GPT-5.4. Max reasoning. Meta attributes the Claude scores to Anthropic system cards.	60.5 points Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort. The source specifies neither a denominator nor a normalized scale. This is the published raw score, not a percentage.	Winner62.3 points Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort. The source specifies neither a denominator nor a normalized scale. This is the published raw score, not a percentage.	Meta Source details Data as of: 10/15/2026
Humanity's Last ExamOverall Test details No documented difference in the test setupVersion: unspecifiedMetric: Text-only pass@1 accuracyScoring: Higher is betterMuse Wonder: Snapshot: Muse Wonder (Max), Reasoning: Max Muse Wonder Full: Snapshot: Muse Wonder Full (Max), Reasoning: MaxMuse Wonder: Exact AA configuration: Muse Wonder (Max) (release muse-wonder, row muse-wonder, UUID ff2f3f3d-783f-4e27-8a39-47237501a313, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.Muse Wonder Full: Exact AA configuration: Muse Wonder Full (Max) (release muse-wonder-full, row muse-wonder-full, UUID e994a6e0-d36b-4dab-8ae1-c1b469e4d9e2, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.	50.1% Artificial Analysis describes this comparable run as 2,158 text-only questions, not the full 2,500-question dataset.	Winner53.6% Artificial Analysis describes this comparable run as 2,158 text-only questions, not the full 2,500-question dataset.	Artificial Analysis Source details Data as of: 10/15/2026
IOIOverall Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	66.944 %	Winner79.833 %	Vals AI Source details Data as of: 10/14/2026
JobBenchOverall Test details No documented difference in the test setupVersion: Meta Wonder scorecard; version unspecifiedMetric: mean task rewardScoring: Higher is betterSettings: Reasoning: max65 professional tasks across 35 occupations. Mean task-specific reward, max reasoning.	70.5 points Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort. The source specifies neither a denominator nor a normalized scale. This is the published raw score, not a percentage.	Winner75.3 points Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort. The source specifies neither a denominator nor a normalized scale. This is the published raw score, not a percentage.	Meta Source details Data as of: 10/15/2026
Legal Research BenchOverall · All-pass Test details No documented difference in the test setupVersion: 1Metric: all-pass rateScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	50.481 %	Winner53.365 %	Vals AI Source details Data as of: 10/12/2026
MemoryBenchOverall Test details No documented difference in the test setupVersion: Meta internal Wonder evaluationMetric: LLM-judge weighted rubric scoreScoring: Higher is betterSettings: Reasoning: maxInternal Meta evaluation with 64K-100K-token conversations, distractors and conflicting or updated turns. Max reasoning.	Winner85.6 points Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort. The source specifies neither a denominator nor a normalized scale. This is the published raw score, not a percentage.	85.5 points Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort. The source specifies neither a denominator nor a normalized scale. This is the published raw score, not a percentage.	Meta Source details Data as of: 10/15/2026
MysteryMechanismOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	27.477 %	Winner35.135 %	Vals AI Source details Data as of: 10/12/2026
OSWorld V2.1 (binary completion)Overall Test details No documented difference in the test setupVersion: V2.1Metric: strict binary task completionScoring: Higher is betterSettings: Reasoning: max108 Ubuntu desktop workflows. Full task completion without partial credit, max reasoning. No binary Opus 5.5 score is published.	34.3 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Winner39.4 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Meta Source details Data as of: 10/15/2026
OSWorld V2.1 (partial reward)Overall Test details No documented difference in the test setupVersion: V2.1Metric: mean per-task partial rewardScoring: Higher is betterSettings: Reasoning: max108 Ubuntu desktop workflows. Partial reward, max reasoning. Meta attributes the Opus 5.5 score to Anthropic.	72.6 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Winner78.4 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Meta Source details Data as of: 10/15/2026
Personal Superintelligence IndexOverall Test details No documented difference in the test setupVersion: Meta Wonder launch scorecardMetric: Meta internal composite scoreScoring: Higher is betterSettings: Reasoning: maxMeta internal composite of seven personal and professional work evaluations. Max reasoning.	54.3 points Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort. The source specifies neither a denominator nor a normalized scale. This is the published raw score, not a percentage.	Winner59.1 points Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort. The source specifies neither a denominator nor a normalized scale. This is the published raw score, not a percentage.	Meta Source details Data as of: 10/15/2026
ProgramBenchOverall Test details No documented difference in the test setupVersion: Meta Wonder scorecard; version unspecifiedMetric: rollouts passing at least 95% of testsScoring: Higher is betterSettings: Reasoning: max, Harness: mini-swe-agentReconstruct programs from a specification. Common mini-swe-agent, clean environment without internet after setup. Share of rollouts passing at least 95% of tests, max reasoning.	50 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Winner59 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Meta Source details Data as of: 10/15/2026
ProofBenchOverall Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	84 %	Winner94 %	Vals AI Source details Data as of: 10/12/2026
Public Benefits Bench v1.1Overall Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	70.23 %	Winner73.748 %	Vals AI Source details Data as of: 10/12/2026
SWE-bench MultimodalOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	45.417 %	Winner47.292 %	Vals AI Source details Data as of: 10/13/2026
SWE-Marathon v1.1Overall Test details No documented difference in the test setupVersion: v1.1Metric: pass@1Scoring: Higher is betterSettings: Reasoning: max, Agent: Muse CodeOfficial Abundant AI leaderboard, max reasoning, Muse Code agent. Long-horizon tasks graded by project tests.	52.5 % The directly retrieved Abundant AI leaderboard reports 53.1% for Full and 52.5% for Wonder, both max with Muse Code. Meta instead reports 57.5% at xhigh for Full and 53.1% at max for Wonder. The runs remain separate; the cause of the difference is unresolved.	Winner53.1 % The directly retrieved Abundant AI leaderboard reports 53.1% for Full and 52.5% for Wonder, both max with Muse Code. Meta instead reports 57.5% at xhigh for Full and 53.1% at max for Wonder. The runs remain separate; the cause of the difference is unresolved.	SWE-Marathon Source details Data as of: 10/15/2026
Tax Agent BenchOverall Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	36.269 %	Winner42.997 %	Vals AI Source details Data as of: 10/12/2026
Terminal-BenchOverall Test details No documented difference in the test setupVersion: 4.0Metric: Task successScoring: Higher is betterMuse Wonder: Snapshot: Muse Wonder (Max), Reasoning: Max Muse Wonder Full: Snapshot: Muse Wonder Full (Max), Reasoning: MaxMuse Wonder: Exact AA configuration: Muse Wonder (Max) (release muse-wonder, row muse-wonder, UUID ff2f3f3d-783f-4e27-8a39-47237501a313, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.Muse Wonder Full: Exact AA configuration: Muse Wonder Full (Max) (release muse-wonder-full, row muse-wonder-full, UUID e994a6e0-d36b-4dab-8ae1-c1b469e4d9e2, reasoning=true, source reasoning-token configuration budget=2000, suffix=Max). This configured value is not measured token consumption.	52.0%	Winner57.1%	Artificial Analysis Source details Data as of: 10/15/2026
Terminal-Bench 4.0Overall Test details No documented difference in the test setupVersion: 4.0Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	49.495 %	Winner58.081 %	Vals AI Source details Data as of: 10/13/2026
Terminal-Bench 4.0-AAOverall Test details No documented difference in the test setupVersion: 4.0-AAMetric: task accuracyScoring: Higher is betterSettings: Reasoning: maxTerminal tasks in isolated environments with executable verifiers. Max reasoning. Meta names Artificial Analysis as the source; each agent remains part of the result.	52 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Winner57.1 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Meta Source details Data as of: 10/15/2026
Terminal-Bench ScienceOverall Test details No documented difference in the test setupVersion: 0.1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Harness: Mini-SWE-agent, Temperature: 1, Output limit: 131,072 tokens	22.857 %	Winner37.143 %	Vals AI Source details Data as of: 10/14/2026
Vals Finance Agent V2Overall Test details No documented difference in the test setupVersion: V2Metric: private test accuracyScoring: Higher is betterSettings: Reasoning: maxFinancial research using filings and market data. Private test set, max reasoning. Vals results republished by Meta.	56.6 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Winner59.1 % Comparison table published by Meta. Depending on the benchmark, Meta uses internal runs, provider reports or third-party results and selects the highest available validated score. This is not a uniformly independently measured cohort.	Meta Source details Data as of: 10/15/2026
Vals IndexOverall Test details No documented difference in the test setupVersion: 2.1Metric: weighted index scoreScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	61.062 %	Winner64.566 %	Vals AI Source details Data as of: 10/13/2026
Vibe Code Bench 1-100Overall Test details No documented difference in the test setupVersion: 1.0Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Harness: OpenHands, Temperature: 1, Output limit: 131,072 tokens	26.55 %	Winner33.978 %	Vals AI Source details Data as of: 10/14/2026
Vibe Code Bench v1.1Overall Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Harness: OpenHands, Temperature: 1, Output limit: 131,072 tokens	83.089 %	Winner85.996 %	Vals AI Source details Data as of: 10/12/2026
BioMysteryBenchHard Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Harness: Terminus 2, Temperature: 1, Output limit: 131,072 tokens	29.412 %	Winner33.333 %	Vals AI Source details Data as of: 10/12/2026
Code MigrationCLI Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	Winner60.07 %	58.044 %	Vals AI Source details Data as of: 10/12/2026
CyberBench v1.1Patch Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	Winner85.714 %	83.929 %	Vals AI Source details Data as of: 10/12/2026
EMBComps Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	79.488 %	Winner80.89 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)Adjustments Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	52.67 %	Winner56.668 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkAntitrust Competition · Criteria passed Test details No documented difference in the test setupVersion: 1Metric: criteria pass rateScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	93.841 %	Winner94.928 %	Vals AI Source details Data as of: 10/12/2026
IOIIOI 2024 Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	66.833 %	Winner75.333 %	Vals AI Source details Data as of: 10/14/2026
Legal Research BenchAdministrative / Regulatory · All-pass Test details No documented difference in the test setupVersion: 1Metric: all-pass rateScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	Tie63.636 %	Tie63.636 %	Vals AI Source details Data as of: 10/12/2026
SWE-bench Multimodalbpmn-js Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	67.925 %	Winner69.811 %	Vals AI Source details Data as of: 10/13/2026
Tax Agent BenchControversy & Precedence Analysis Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	20 %	Winner26.667 %	Vals AI Source details Data as of: 10/12/2026
Terminal-Bench 4.0Hardware Test details No documented difference in the test setupVersion: 4.0Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	Tie46.667 %	Tie46.667 %	Vals AI Source details Data as of: 10/13/2026
Terminal-Bench ScienceEarth Sciences Test details No documented difference in the test setupVersion: 0.1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Harness: Mini-SWE-agent, Temperature: 1, Output limit: 131,072 tokens	50 %	Winner62.5 %	Vals AI Source details Data as of: 10/14/2026
Vals IndexCode Migration Test details No documented difference in the test setupVersion: 2.1Metric: benchmark scoreScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	Winner62.009 %	60.256 %	Vals AI Source details Data as of: 10/13/2026
BioMysteryBenchHuman-solvable Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Harness: Terminus 2, Temperature: 1, Output limit: 131,072 tokens	75.799 %	Winner82.648 %	Vals AI Source details Data as of: 10/12/2026
Code MigrationCOBOL Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	Tie70 %	Tie70 %	Vals AI Source details Data as of: 10/12/2026
CyberBench v1.1PoC Test details No documented difference in the test setupVersion: 1.1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	Winner75 %	63.333 %	Vals AI Source details Data as of: 10/12/2026
EMBDataroom Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	79.232 %	Winner85.257 %	Vals AI Source details Data as of: 10/12/2026
Finance Agent (v2)All-Pass Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	45.936 %	Winner49.037 %	Vals AI Source details Data as of: 10/12/2026
Harvey's Legal Agent BenchmarkAntitrust Competition · Task fully resolved Test details No documented difference in the test setupVersion: 1Metric: task resolution rateScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	Tie20 %	Tie20 %	Vals AI Source details Data as of: 10/12/2026
IOIIOI 2025 Test details No documented difference in the test setupVersion: 2Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	75.5 %	Winner85.5 %	Vals AI Source details Data as of: 10/14/2026
Legal Research BenchAdministrative / Regulatory · Weighted partial credit Test details No documented difference in the test setupVersion: 1Metric: weighted partial-credit scoreScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	90.571 %	Winner92.809 %	Vals AI Source details Data as of: 10/12/2026
SWE-bench MultimodalCarbon Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	27.068 %	Winner29.323 %	Vals AI Source details Data as of: 10/13/2026
Tax Agent BenchControversy & Precedence Analysis Test details No documented difference in the test setupVersion: 1Metric: accuracyScoring: Higher is betterSettings: Reasoning: max, Temperature: 1, Output limit: 131,072 tokens	71.362 %	Winner76.097 %	Vals AI Source details Data as of: 10/12/2026
Key to the marks:better valuebetter, but not conclusivetie

Our model tests

Our tests

Compare how the models respond to the same prompt. Each test shows the latest attempt, with no subsequent fixes to the generated code. Earlier runs remain available in the test details. These results do not contribute to an overall score.

Pelican on a bicycle

A community classic for free-form SVG drawing.

Muse Wonder

Not tested yet

Muse Wonder Full

Not tested yet

Prompt and test conditions

Generate an SVG of a pelican riding a bicycle

Token limit including reasoning: 8,192 tokens

Task origin (Simon Willison)

How to read this comparison

Winning one benchmark is not an overall verdict

Your workload, budget, and required context length matter most. A coding benchmark says little about visual understanding or agent performance.

The standard price table uses direct developer rates. The provider comparison labels OpenRouter endpoints, regions, and special tariffs separately.

Every result links to its measurement source and retrieval date. Labels distinguish vendor reports, official benchmark leaderboards, and independent evaluations. Disputed or archived results are excluded.

More matchups

Compare Muse Wonder and Muse Wonder Full with other leading models using the same data and benchmark logic.

View the complete LLM comparison