Muse Spark vs Muse Wonder Full
Full profiles: Muse Spark (Meta) and Muse Wonder Full (Meta): every benchmark, price and rank.
Updated · Data from 7 sources
As of 10 Nov 2026, Muse Spark and Muse Wonder Full are neck and neck on quality (64.2 vs 62.2 on the BenchLeader Index, inside the ±4.2 margin). Choose Spark for reasoning (67 vs 66); choose Wonder Full for coding (70 vs 65) and long context (67 vs 64).
Key facts (as of 10 Nov 2026)
- Muse Spark scores 64.2 on the BenchLeader Index and Muse Wonder Full scores 62.2 (BenchLeader, 10 Nov 2026).
- Of the 8 benchmarks both models have, Muse Wonder Full leads 7 (BenchLeader, 10 Nov 2026).
Which to pick for…
The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.
| Use case | Pick | Why | Confidence |
|---|---|---|---|
| Coding | Muse Wonder Full | 70 vs 65 category score; Vibe Code Bench v1.1 86.0% vs 19.7% | Medium |
| Agents & tool use | Either | 65 vs 66 category score | Either |
| Reasoning | Muse Spark | 67 vs 66 category score; Humanity's Last Exam (AA) 40.7% vs 53.6% | Medium |
| Knowledge | Either | 63 vs 63 category score; LegalBench 84.2% vs 87.4% | Either |
| Long documents | Muse Wonder Full | 67 vs 64 category score; AA-LCR v1.1 83.7% vs 78.0% | Medium |
| Chat quality (human preference) | – | Only Muse Spark has an Arena rating (68 category score) | Not comparable |
| High volume on a budget | Neither flagship | A smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs | – |
Key differences
- Muse Spark and Muse Wonder Full (max) are neck and neck on quality (64.2 vs 62.2, within the ±4.2 margin).
- Muse Spark is stronger in human preference, instruction following, knowledge, maths, multimodal, reasoning.
- Muse Wonder Full (max) is stronger in agents & tools, coding, long context.
At a glance
| Spec | Muse Spark | Muse Wonder Full |
|---|---|---|
| Developer | Meta | Meta |
| Released | 8 Apr 2026 | 12 Oct 2026 |
| Context window (tokens) | 262,144 | 1,000,000 |
| Compared at | Muse Spark | Muse Wonder Full (max) |
| Open weights | No | No |
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 | Muse Spark | Muse Wonder Full (max) | Leader |
|---|---|---|---|
| Reasoning | 67.4 | 66.2 | Muse Spark |
| Coding | 65.2 | 70.3 | Muse Wonder Full |
| Agents & tools | 65.0 | 65.8 | level |
| Maths | 59.2 | – | not comparable |
| Knowledge | 63.1 | 62.6 | level |
| Instruction following | 76.9 | – | not comparable |
| Human preference | 68.1 | – | not comparable |
| Multimodal | 64.2 | – | not comparable |
| Long context | 63.6 | 66.6 | Muse Wonder Full |
On one chart
Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.
Not drawn, measured for fewer than two of these models: Maths, Instruction following, Multimodal, Human preference.
Where sources disagree
Across the 8 benchmarks both have, Muse Spark leads 1, Muse Wonder Full leads 7.
| Overall index | Muse Spark | Muse Wonder Full (max) | Leader |
|---|---|---|---|
| BenchLeader Index | 64.2 | 62.2 | neck and neck |
Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).
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. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.
| Measure | Muse Spark | Muse Wonder Full (max) |
|---|---|---|
| Output tokens per task (AA Index) | – | 88k |
| Of which reasoning | – | 55k |
Each model is compared at its best-ranked reasoning-effort setting. Muse Spark scores 64.2. Muse Wonder Full (max) scores 62.2. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.
Every benchmark
101 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.
| Metric | Publisher | Muse Spark | Muse Wonder Full (max) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 64.2 | 62.2 |
| Agents & tools score | BenchLeader | 65.0 | 65.8 |
| Coding score | BenchLeader | 65.2 | 70.3 |
| Human preference score | BenchLeader | 68.1 | – |
| Instruction following score | BenchLeader | 76.9 | – |
| Knowledge score | BenchLeader | 63.1 | 62.6 |
| Long context score | BenchLeader | 63.6 | 66.6 |
| Maths score | BenchLeader | 59.2 | – |
| Multimodal score | BenchLeader | 64.2 | – |
| Reasoning score | BenchLeader | 67.4 | 66.2 |
| Context window | BenchLeader | 262,144 tokens | 1,000,000 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 89.8% | – |
| Humanity's Last Exam | Scale AI / CAIS | 40.6% | – |
| Arena Hard Prompts | Arena (LMArena) | 1506 | – |
| GPQA Diamond (AA) | Artificial Analysis | 88.4% | – |
| Humanity's Last Exam (AA) | Artificial Analysis | 40.7% | 53.6% |
| HLE-Diamond (AA) | Artificial Analysis | – | 48.5% |
| GPQA Diamond (Vals) | Vals AI | 89.7% | – |
| CritPt | Artificial Analysis | 11.3% | 32.9% |
| CritPt v2 hard subset (AA) | Artificial Analysis | – | 56.5% |
| MysteryMechanism | Vals AI | – | 35.1% |
| Coding | |||
| SciCode | SciCode | 51.5% | – |
| Arena Coding | Arena (LMArena) | 1529 | – |
| Arena WebDev | Arena (LMArena) | – | 1724 |
| SciCode (AA) | Artificial Analysis | – | 62.6% |
| IOI | Vals AI | – | 79.8% |
| SWE-bench (Vals) | Vals AI | 74.4% | – |
| SWE-Bench Pro | Scale AI SEAL | 55.0% | – |
| Code Migration | Vals AI | – | 61.0% |
| Code Migration v1.1 | Vals AI | – | 62.8% |
| ProgramBench | Vals AI | – | 0.5% |
| SWE-bench Multimodal | Vals AI | – | 47.3% |
| Vibe Code Bench 1-100 | Vals AI | – | 34.0% |
| Vibe Code Bench v1.1 | Vals AI | 19.7% | 86.0% |
| SWE-Bench Pro (private) | Scale AI SEAL | 55.0% | – |
| SWE Atlas: Codebase QnA | Scale AI SEAL | 24.2% | – |
Show all 101 benchmarks
| Metric | Publisher | Muse Spark | Muse Wonder Full (max) |
|---|---|---|---|
| Coding | |||
| SWE Atlas: Test Writing | Scale AI SEAL | 31.1% | – |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | – | 57.1% |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | 62.2% | – |
| Design Arena Website | Design Arena | 1257 | – |
| FrontierCode 1.1 (AA) | Artificial Analysis | – | 44.9% |
| FrontierCode 2.0 (AA) | Artificial Analysis | – | 37.8% |
| Agents & tools | |||
| Arena Agent | Arena (LMArena) | – | 9.1 |
| Terminal-Bench Hard | Artificial Analysis | 45.5% | – |
| τ²-Bench Telecom (AA) | Artificial Analysis | 91.5% | – |
| MCP Atlas | Scale AI SEAL | 82.2% | – |
| GDPval-AA v2.1, normalised score | Artificial Analysis | – | 67.8% |
| GDPval-AA v2.1 | Artificial Analysis | – | 1857 |
| CyberBench | Vals AI | – | 73.6% |
| Finance Agent v2 | Vals AI | – | 59.1% |
| Finance Agent v2.1 | Vals AI | – | 67.6% |
| Harvey's Legal Agent Benchmark v1.1 | Vals AI | – | 23.3% |
| Harvey's Legal Agent Benchmark v1 | Vals AI | – | 21.7% |
| Legal Research Bench | Vals AI | – | 53.4% |
| Robo-bench | Vals AI | – | 10.4% |
| SREBench | Vals AI | – | 11.8% |
| Tax Agent Bench | Vals AI | – | 43.0% |
| Terminal-Bench 2.0 (Vals) | Vals AI | 59.5% | – |
| Terminal-Bench 4.0 (Vals) | Vals AI | – | 58.1% |
| Terminal-Bench Science | Vals AI | – | 37.1% |
| AutomationBench | Artificial Analysis | – | 71.3% |
| GDP.pdf (AA) | Artificial Analysis | – | 32.6% |
| Harvey LAB-AA v1.1 | Harvey | – | 13.1% |
| AA-Briefcase v1.1 | Artificial Analysis | – | 1729 |
| GDPx | GDPx by SemiAnalysis | – | 9.0% |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | – | 37.1% |
| Maths | |||
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 88.9% | – |
| ProofBench | Vals AI | 17.0% | – |
| AIME (Vals) | Vals AI | 96.9% | – |
| Arena Maths | Arena (LMArena) | 1469 | – |
| Knowledge | |||
| AA-Omniscience v1.1 | Artificial Analysis | – | 30.6 |
| MMLU-Pro | Vals AI | 87.3% | – |
| LegalBench | Vals AI | 84.2% | 87.4% |
| CorpFin | Vals AI | 65.1% | – |
| TaxEval | Vals AI | 77.7% | – |
| PRBench Finance | Scale AI SEAL | 52.4% | – |
| PRBench Legal | Scale AI SEAL | 52.3% | – |
| MultiNRC | Scale AI SEAL | 59.0% | – |
| BioMysteryBench | Vals AI | – | 73.3% |
| CaseLaw v2 | Vals AI | 63.1% | – |
| Excel Modeling Benchmark | Vals AI | – | 71.9% |
| Excel Modeling Benchmark v1.1 | Vals AI | – | 75.0% |
| MedCode | Vals AI | 51.3% | 49.9% |
| MedScribe | Vals AI | 85.9% | 89.3% |
| Public Benefits Bench v1.1 | Vals AI | – | 73.8% |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | – | 41.0% |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | – | 82.4% |
| Instruction following | |||
| IFBench | Artificial Analysis | 75.9% | – |
| MultiChallenge | Scale AI SEAL | 75.5% | – |
| TutorBench | Scale AI SEAL | 68.5% | – |
| Arena Instruction Following | Arena (LMArena) | 1465 | – |
| Human preference | |||
| Arena Text | Arena (LMArena) | 1489 | – |
| Arena Creative Writing | Arena (LMArena) | 1464 | – |
| Arena Multi-turn | Arena (LMArena) | 1493 | – |
| Arena Longer Queries | Arena (LMArena) | 1475 | – |
| Multimodal | |||
| Arena Vision | Arena (LMArena) | 1306 | – |
| MMMU-Pro | Artificial Analysis | 80.5% | – |
| MMMU-Pro (official) | MMMU | 80.4%† | – |
| MMMU-Pro (Vals) | Vals AI | 87.4% | – |
| Long context | |||
| AA-LCR v1.1 | Artificial Analysis | 78.0% | 83.7% |
| Arena Document | Arena (LMArena) | 1467 | – |
| MLCR | Artificial Analysis | – | 22.2% |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 153.1 | – |
| AA Intelligence Index v5.0 | Artificial Analysis | – | 50.1 |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 31.3 | 55.8 |
| AA Progress Index v1 | Artificial Analysis | 289.1 | 343.2 |
| Vals Index | Vals AI | – | 66.8 |
| Vals Finance Index | Vals AI | – | 64.5% |
| Vals RSI Index | Vals AI | – | 24.4% |
| Vals Science Index | Vals AI | – | 48.5% |
| Vals Router Index | Vals AI | – | 66.8%† |
| Safety & honesty | |||
| FORTRESS | Scale AI SEAL | 20.2% | – |
Data as of 2026-11-10. Best configuration of each model; every score links to its source on the model pages.
What changed
- Muse Spark: BenchLeader Index 64.6 → 64.1.
- Muse Spark: 1 score revision.
- Muse Wonder Full (max): 1 score revision.
- Muse Wonder Full (max): 1 score revision.
- Muse Wonder Full (max): 1 new result (Arena Agent).
- Muse Spark: 1 new result (AA Progress Index v1).
- Muse Wonder Full (max): 6 new results (AA Intelligence Index v5.0, AA Progress Index v1, HLE-Diamond (AA)…); 2 score revisions.
- Muse Wonder Full (max): 1 new result (Terminal-Bench-Science 0.1 (AA)).
- Muse Wonder Full (max): 9 new results (Code Migration v1.1, Excel Modeling Benchmark v1.1, Finance Agent v2.1…).
- Muse Wonder Full (max): 1 new result (FrontierCode 1.1 (AA)).
Related comparisons
- Muse Spark vs Claude Fable 5.2
- Muse Spark vs Claude Opus 5.5
- Muse Spark vs Gemini 4 Carbon
- Muse Wonder Full vs Claude Fable 5.2
- Muse Wonder Full vs Claude Opus 5.5
- Muse Wonder Full vs Gemini 4 Carbon
- Muse Wonder Full vs GPT-6 Astra
- Muse Spark vs Muse Spark 1.3
Rankings: Best LLM overall · Best LLM for coding · Best LLM for agents · Best cheap LLMs
Muse Spark vs Muse Wonder Full: questions
- Is Muse Spark better than Muse Wonder Full?
- Muse Spark and Muse Wonder Full (max) are neck and neck on quality (64.2 vs 62.2, within the ±4.2 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 10 Nov 2026, so check the category scores for your use.
- Is Muse Spark or Muse Wonder Full better for coding?
- Muse Wonder Full scores higher in coding (70 vs 65 on the category index, where 50 is average).
- Is Muse Spark or Muse Wonder Full better for agents and tool use?
- They are level in agentic tasks (65 vs 66 on the category index).
- Which is better for long documents?
- Muse Wonder Full scores higher on long-context benchmarks (67 vs 64). Muse Wonder Full accepts more context: 1,000,000 against 262,144 tokens for Muse Spark.
- Do independent benchmarks agree on Muse Spark vs Muse Wonder Full?
- Across the 8 benchmarks both have, Muse Spark leads 1, Muse Wonder Full leads 7.
More questions (1)
- When was Muse Wonder Full released?
- Muse Wonder Full was released on 12 Oct 2026.