GPT-6 Sol vs Muse Wonder Full
As of 18 Oct 2026, Muse Wonder Full (max) is the better pick for most workloads: it leads on quality (68.6 vs 65.6 on the BenchLeader Index).
GPT-6 Sol 
Muse Wonder Full Key differences
- Muse Wonder Full (max) leads on quality: 68.6 vs 65.6.
- GPT-6 Sol (max) is stronger in agents & tools, coding, human preference, maths, multimodal.
- Muse Wonder Full (max) is stronger in composite, knowledge, reasoning.
Choose GPT-6 Sol (max) if
- coding is the priority (67 vs 64 on the category score)
Choose Muse Wonder Full (max) if
- reasoning is the priority (95 vs 72 on the category score)
- knowledge is the priority (71 vs 70 on the category score)
Category by category
Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.
| Category | GPT-6 Sol (max) | Muse Wonder Full (max) | Leader |
|---|---|---|---|
| Reasoning | 71.5 | 95.0 | Muse Wonder Full |
| Coding | 67.4 | 64.0 | GPT-6 Sol |
| Agents & tools | 60.3 | – | not comparable |
| Maths | 69.3 | – | not comparable |
| Knowledge | 69.6 | 71.3 | Muse Wonder Full |
| Human preference | 64.9 | – | not comparable |
| Multimodal | 61.9 | – | not comparable |
| Long context | 67.0 | 67.0 | level |
| Composite | 72.6 | 92.2 | Muse Wonder Full |
Price, speed and latency
List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page; suite cost is what Artificial Analysis paid to run its whole Intelligence Index.
| Measure | GPT-6 Sol (max) | Muse Wonder Full (max) |
|---|---|---|
| Blended price per 1M tokens | $4.00 | – |
| Input / output per 1M | $2.00 / $10.00 | – / – |
| Price per task: chat reply | $0.0038 | – |
| Price per task: summarise a 30-page report | $0.030 | – |
| Price per task: code edit | $0.027 | – |
| Price per task: agentic coding session | $0.160 | – |
| Price per task: structured extraction | $0.0060 | – |
| Cost to run a full benchmark suite | $1.04 | – |
| Output speed | 113 tok/s | – |
| Time to first answer | 115.3 s | – |
| Time to a complete answer | 119.7 s | – |
Every benchmark
| Metric | GPT-6 Sol (max) | Muse Wonder Full (max) |
|---|---|---|
| BenchLeader Index | 65.6 | 68.6 |
| Agents & tools score | 60.3 | – |
| Coding score | 67.4 | 64.0 |
| Composite score | 72.6 | 92.2 |
| Human preference score | 64.9 | – |
| Knowledge score | 69.6 | 71.3 |
| Long context score | 67.0 | 67.0 |
| Maths score | 69.3 | – |
| Multimodal score | 61.9 | – |
| Reasoning score | 71.5 | 95.0 |
| Blended price $/M | $4.00 | – |
| Output speed | 113 tok/s | – |
| Time to first answer | 115.3 s | – |
| Context window | 1.1M | 1M |
| GPQA Diamond | 94.3% | – |
| FrontierMath Tiers 1–3 | 89.8% | – |
| FrontierMath Tier 4 | 90.0% | – |
| OTIS Mock AIME | 100.0% | – |
| SimpleQA Verified | 60.7% | – |
| Terminal-Bench | 49.4% | – |
| SciCode | 57.6% | – |
| APEX-Agents | 54.3% | – |
| FrontierCode | 49.3% | – |
| Arena Text | 1459 | – |
| Arena Hard Prompts | 1484 | – |
| Arena Coding | 1518 | – |
| Arena WebDev | 1683 | – |
| Arena Vision | 1245 | – |
| Arena Agent | 10.3 | – |
| LiveBench | 79.3% | – |
| LiveBench Reasoning | 88.7% | – |
| LiveBench Coding | 81.8% | – |
| LiveBench Agentic Coding | 52.9% | – |
| LiveBench Mathematics | 96.4% | – |
| LiveBench Data Analysis | 81.2% | – |
| LiveBench Language | 85.3% | – |
| LiveBench Instruction Following | 68.6% | – |
| AA Intelligence Index v4.3.2 | 47.6 | 55.8 |
| AA-LCR | 83.7% | 83.7% |
| MMMU-Pro | 83.0% | – |
| AA-Omniscience | 27.1 | 30.1 |
| Humanity's Last Exam (AA) | 47.9% | 53.6% |
| SciCode (AA) | 57.6% | 62.6% |
| IOI | 82.6% | 79.8% |
| LegalBench | – | 87.4% |
| Terminal-Bench 2.1 (Vals) | 83.2% | – |
| Vals Index | 57.5 | 64.6 |
| ARC-AGI-1 | 95.5% | – |
| ARC-AGI-2 | 89.6% | – |
| ARC-AGI-3 | 23.0% | – |
| CritPt | 30.9% | 32.9% |
| GDPval-AA v2.1 | 50.4% | 67.8% |
| ITBench SRE (AA) | 49.4% | – |
| BioMysteryBench | 74.8% | 73.3% |
| Code Migration | 57.2% | 61.0% |
| CyberBench | 78.0% | 73.6% |
| Excel Modeling Benchmark | 71.5% | 71.9% |
| Finance Agent v2 | 49.0% | 59.1% |
| Harvey's Legal Agent Benchmark | 1.7% | 21.7% |
| Legal Research Bench | 28.9% | 53.4% |
| MedCode | 47.1% | 49.9% |
| MedScribe | 82.0% | 89.3% |
| MysteryMechanism | 30.2% | 35.1% |
| ProgramBench | 2.0% | 0.5% |
| Public Benefits Bench | 56.6% | 73.8% |
| SAGE | 44.8% | – |
| SREBench | 30.5% | 11.8% |
| Tax Agent Bench | 15.0% | 43.0% |
| Terminal-Bench 4.0 (Vals) | 44.4% | 58.1% |
| Terminal-Bench Science | 30.0% | 37.1% |
| Vibe Code Bench 1-100 | – | 34.0% |
| Vibe Code Bench v1.1 | 87.8% | 86.0% |
| Arena Maths | 1450 | – |
| Arena Creative Writing | 1438 | – |
| Arena Instruction Following | 1452 | – |
| Arena Multi-turn | 1467 | – |
| Arena Longer Queries | 1465 | – |
| Arena Search | 1228 | – |
| Chess Puzzles | 56.0% | – |
| EBR-bench | 53.3% | – |
| Mystery Game Puzzles | 56.0% | – |
| LMCA | 59.1% | – |
| DTBench | 97.3% | – |
| ALE-Bench | 2462 | – |
| GDP.pdf | 26.4% | – |
| Terminal-Bench 4.0 (AA) | 43.9% | 57.1% |
| AutomationBench | 61.6% | 71.3% |
| GDP.pdf | 25.2% | 32.6% |
| MLCR | 16.1% | – |
| Harvey LAB-AA v1.1 | 3.6% | 13.1% |
| AA-Omniscience: accuracy | 54.5% | 43.2% |
| AA-Omniscience: non-hallucination | 39.9% | 77.0% |
| AA-Briefcase v1.1 | 1481 | 1729 |
| Terminal-Bench-Science 0.1 (AA) | 30.0% | – |
| AA Cyber Index | 3682.8% | – |
| CWE-Bench (AA) | 64.2% | – |
| DeepSec-Bench (AA) | 46.3% | – |
| CyberGym end-to-end (AA) | 0.0% | – |
| AA-3D | 1123 | – |
Data as of 2026-10-18. Best configuration of each model; every score links to its source on the model pages.
GPT-6 Sol vs Muse Wonder Full: questions
- Is GPT-6 Sol better than Muse Wonder Full?
- Muse Wonder Full (max) leads on quality: 68.6 vs 65.6. The BenchLeader Index combines every independent quality benchmark; Muse Wonder Full (max) is ahead overall as of 2026-10-18, but check the category scores for your use.
- Is GPT-6 Sol better than Muse Wonder Full for coding?
- GPT-6 Sol scores higher in coding (67 vs 64 on the category index, where 50 is average).
- Which has the larger context window?
- GPT-6 Sol accepts more context: 1.1M against 1M tokens.