GPT-5.6 Luna vs Muse Spark 1.1
GPT-5.6 Luna 
Muse Spark 1.1 Verdict
- Muse Spark 1.1 leads on quality: 65.0 vs 59.4.
- GPT-5.6 Luna (xhigh) is stronger in composite, long context.
- Muse Spark 1.1 is stronger in agents & tools, coding, human preference, knowledge, multimodal, reasoning, instruction following, maths.
- GPT-5.6 Luna (xhigh) is 4.4× cheaper ($0.450 vs $2.00 per 1M blended).
- Muse Spark 1.1 streams 1.8× faster (194 vs 107 tokens per second).
| Metric | GPT-5.6 Luna (xhigh) | Muse Spark 1.1 |
|---|---|---|
| BenchLeader Index | 59.4 | 65.0 |
| Agents & tools score | 50.1 | 62.7 |
| Coding score | 60.0 | 68.1 |
| Composite score | 72.9 | 72.3 |
| Human preference score | 65.0 | 66.3 |
| Knowledge score | 59.0 | 73.2 |
| Long context score | 67.4 | 65.4 |
| Multimodal score | 61.6 | 65.0 |
| Reasoning score | 53.9 | 69.2 |
| Instruction following score | – | 77.8 |
| Maths score | – | 52.2 |
| Blended price $/M | $0.450 | $2.00 |
| Output speed | 107 tok/s | 194 tok/s |
| Time to first answer | 55.4 s | 2.6 s |
| Context window | 1M | 1.0M |
| SimpleQA Verified | – | 57.8% |
| SimpleBench | 46.8% | – |
| SciCode | 50.0% | 58.2% |
| APEX-Agents | – | 41.9% |
| ProofBench | – | 39.0% |
| Epoch Capabilities Index | – | 154.6 |
| LMArena Text | 1453 | 1492 |
| LMArena Hard Prompts | 1473 | 1511 |
| LMArena Coding | 1497 | 1531 |
| LMArena WebDev | 1520 | 1541 |
| LMArena Vision | 1254 | 1293 |
| LMArena Agent | 1.1 | -2.6 |
| AA Intelligence Index | 34.8 | 34.3 |
| AA-LCR | 81.7% | 77.7% |
| MMMU-Pro | 78.5% | – |
| AA-Omniscience | -10.8 | 28.1 |
| GPQA Diamond (AA) | 89.5% | 89.8% |
| Humanity's Last Exam (AA) | 37.0% | 46.2% |
| SciCode (AA) | 50.5% | 58.8% |
| SWE-Bench Pro | – | 61.5% |
| MCP Atlas | – | 88.1% |
| MultiChallenge | – | 75.3% |
| PRBench Finance | – | 55.0% |
| PRBench Legal | – | 57.0% |
| MultiNRC | – | 65.6% |
| EQ-Bench 4 | – | 1260 |
| ARC-AGI-1 | 87.7% | – |
| ARC-AGI-2 | 47.6% | – |
| ARC-AGI-3 | 0.0% | – |
Data as of 2026-09-10. Best configuration of each model; every score links to its source on the model pages.
GPT-5.6 Luna vs Muse Spark 1.1: questions
- Is GPT-5.6 Luna better than Muse Spark 1.1?
- Muse Spark 1.1 leads on quality: 65.0 vs 59.4. The BenchLeader Index combines every independent quality benchmark; Muse Spark 1.1 is ahead overall as of 2026-09-10, but check the category scores for your use.
- Is GPT-5.6 Luna better than Muse Spark 1.1 for coding?
- Muse Spark 1.1 scores higher in coding (68 vs 60 on the category index, where 50 is average).
- Is GPT-5.6 Luna better than Muse Spark 1.1 for agentic tasks?
- Muse Spark 1.1 scores higher in agentic tasks (63 vs 50 on the category index, where 50 is average).
- Which is cheaper, GPT-5.6 Luna or Muse Spark 1.1?
- GPT-5.6 Luna is cheaper: $0.450 against $2.00 per million tokens, blended at three input tokens per output token.
- Which is faster, GPT-5.6 Luna or Muse Spark 1.1?
- Muse Spark 1.1 streams faster: 194 against 107 output tokens per second.
- Which has the larger context window?
- Muse Spark 1.1 accepts more context: 1.0M against 1M tokens.