Meta•Rank #9NEW

Muse Wonder

Proprietarygeneralcodingagentfast

Leaderboard Score

60.8

Scale: 0 - 100

Throughput Speed

145 tps

Context Window

256K

API Pricing / 1M

.61

Parameter Scale

Proprietary

Monthly API Cost Calculator

Muse Wonder rates: $0.35/1M input • $1.4/1M output. Pick a workload preset or enter your own token volumes.

Monthly input (M tokens)Monthly output (M tokens)

Same workload on published-rate alternatives

Mistral Large 4 (Mistral AI)

Qwen3.5-397B-A17B (Qwen / Alibaba)

Open the comparison engine →

Estimates use listed API rates at time of writing; providers may differ in caching discounts, rate limits, and batching pricing.

Sourced AI Radar & Benchmarks•8 Evaluated Metrics

Capability Radar & Benchmark Scores

Multi-Axis Capability Radar

Comparing Muse Wonder against Frontier Top 10 & Open-Source average

0 - 100 Scale

Muse Wonder

Frontier Top 10 Avg

Open-Source Avg

Performance Vectors

Where Muse Wonder Leads

Coding & SWETop 10: 80% • OS: 51.1%

78.8%-1.2% vs Top 10

Reasoning & MathTop 10: 89.2% • OS: 52.7%

92.4%+3.2% vs Top 10

Agent & Tool AutomationTop 10: 78.3% • OS: 36.4%

72.6%-5.7% vs Top 10

Terminal & CLI ExecutionTop 10: 66.8% • OS: 46.2%

52.0%-14.8% vs Top 10

MMLU & Knowledge QATop 10: 50% • OS: 86.7%

76.0%+26.0% vs Top 10

Cyber & Security ExploitsTop 10: 86.2% • OS: 81.2%

57.8%-28.4% vs Top 10

All Verified Benchmark Evaluations

Direct Pass@1 and resolution rates across standardized evaluations.

OverallCodingReasoningAgenticTerminal

Featured Compute Platform

Run Benchmarks on Dedicated AI Clusters

Zero-cold-start inference endpoints with dedicated GPU instances.

Compare With Alternatives →

Executive Analysis & Verdict

Muse Wonder by Meta is an agile Personal Superintelligence model ranking #8 globally with a 60.8 score, 78.8% DeepSWE 1.1, and 145 tokens per second.

Key Strengths & Highlights

•Exceptional coding accuracy with 78.8% on DeepSWE 1.1 and 53.1% on SWE-Marathon
•High-speed inference throughput of 145 tokens/sec with active memory persistence
•Strong legal reasoning (10.0 on Harvey LAB-AA) and financial analysis (56.6% on Vals Finance)

Limitations & Considerations

•Output token window capped at 32K tokens for single-turn code generation

Recommended Workflows

High-speed autonomous coding agents, interactive terminal copilots, and multi-step desktop workflows

Architectural Overview & Capability Profile

Muse Wonder is Meta AI’s high-speed personal superintelligence model engineered for low-latency coding assistance, real-time agent loops, and persistent workflow automation. Operating with Meta’s proprietary Active Memory and adaptive reasoning engines, Muse Wonder registers an overall RankLLMs score of 60.8 / 100, placing it solidly at #8 globally.

On the independent Artificial Analysis Intelligence Index v5.0, Muse Wonder scores 52, providing frontier-tier intelligence at more than 3x the token throughput of traditional heavy models.

Benchmark Performance Highlights

Software Engineering:
- DeepSWE 1.1: Delivers a remarkable 78.8%, surpassing heavier flagship models like Claude Opus 5.5 (74.2%) and GPT-6 Astra (74.1%).
- SWE-Marathon: Achieves 53.1%, maintaining goal continuity through complex bug fixing and multi-file refactoring.
- Terminal-Bench 4.0-AA: Scores 52.0%, adeptly diagnosing terminal command failures and build scripts.
Professional Knowledge & Superintelligence:
- Personal Superintelligence Index (PSI): Scores 54.3, closely matching GPT-6 Astra (54.5) while running at near-flash speeds.
- Harvey LAB-AA: Scores 10.0, outperforming GPT-6 Astra (8.6) and Claude Opus 5.5 (4.2).
- MemoryBench: Reaches 85.6%, maintaining high contextual fidelity across extended agent sessions.
Computer Use:
- OSWorld V2.1: Achieves 72.6% partial completion, reliably operating GUI software and desktop workflows.

Economics & Speed

Muse Wonder is priced at $0.35 per 1M input tokens and $1.40 per 1M output tokens (blended ~$0.61 / 1M tokens) with an average generation speed of 145 tokens per second. This makes it an ideal workhorse model for production agent swarms and interactive coding environments.

Frequently Asked Questions About Muse Wonder

What is Muse Wonder and who developed it?↓

Muse Wonder is a proprietary large language model developed by Meta. It ranks #9 on the RankLLMs composite leaderboard, ahead of 92% of the 95 models we track, with an overall score of 60.8 / 100.

What context window and inference speed does Muse Wonder offer?↓

Muse Wonder supports a 256K-token input context window and generates about 145 tps tokens per second in our reference measurements and suitable for extended reasoning sessions and multi-turn agent workflows..

How much does Muse Wonder cost per million tokens?↓

API pricing for Muse Wonder is .61 per 1M tokens - cheaper than 58% of priced models we track. Confirm the current rate on the provider's official pricing page before budgeting.

What is Muse Wonder best at?↓

Muse Wonder is strongest in Exceptional coding accuracy with 78.8% on DeepSWE 1.1 and 53.1% on SWE-Marathon; High-speed inference throughput of 145 tokens/sec with active memory persistence; Strong legal reasoning (10.0 on Harvey LAB-AA) and financial analysis (56.6% on Vals Finance). Full pillar-by-pillar numbers are in the benchmark tables above.

What should you watch out for with Muse Wonder?↓

Known trade-offs include: Output token window capped at 32K tokens for single-turn code generation. Weigh these against the strengths above for your specific workload.