GPT-5.6 Terra vs Muse Spark 1.3
Full profiles: GPT-5.6 Terra (OpenAI) and Muse Spark 1.3 (Meta): every benchmark, price and rank.
Updated · Data from 9 sources
As of 24 Oct 2026, GPT-5.6 Terra (max) and Muse Spark 1.3 (max) are neck and neck on quality (64.4 vs 67.1 on the BenchLeader Index, inside the ±6.7 margin of error), and Muse Spark 1.3 (max) costs 2.3× less, which makes it the better buy for most workloads. Pick GPT-5.6 Terra for maths (69 vs 63); pick Muse Spark 1.3 for knowledge (73 vs 57).
GPT-5.6 Terra benchmarks 
Muse Spark 1.3 benchmarks 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 | Either | 65 vs 65 category score; SciCode 55.0% vs 58.8% | Either |
| Agents & tool use | Muse Spark 1.3 | 66 vs 59 category score; τ³-Banking (AA) 50.5% vs 40.2% | Medium |
| Maths | GPT-5.6 Terra | 69 vs 63 category score; FrontierMath Tiers 1–3 86.0% vs 74.0% | Medium |
| Reasoning | Muse Spark 1.3 | 77 vs 70 category score; CritPt 24.9% vs 30.0% | Medium |
| Knowledge | Muse Spark 1.3 | 73 vs 57 category score; AA-Omniscience 25 vs 0.1 | High |
| Long documents | Either | 67 vs 67 category score; AA-LCR 83.0% vs 83.0% | Either |
| Images and other media | Muse Spark 1.3 | 66 vs 64 category score | Medium |
| Chat quality (human preference) | – | Only Muse Spark 1.3 has an Arena rating (69 category score) | Not comparable |
| Lowest cost | Muse Spark 1.3 | $2.00 vs $4.50 per 1M blended; agentic coding session $0.092 vs $0.168 | – |
| Fastest first answer | Either | 91.6 s (GPT-5.6 Terra (max)) vs 90.1 s (Muse Spark 1.3 (max)) to the first answer token | – |
| High volume on a budget | Neither flagship | A smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs | – |
Key differences
- GPT-5.6 Terra (max) and Muse Spark 1.3 (max) are neck and neck on quality (64.4 vs 67.1, within the ±6.7 margin).
- GPT-5.6 Terra (max) is stronger in coding, instruction following, maths.
- Muse Spark 1.3 (max) is stronger in agents & tools, composite, knowledge, multimodal, reasoning, human preference.
- Muse Spark 1.3 (max) is 2.3× cheaper ($2.00 vs $4.50 per 1M blended).
At a glance
| Spec | GPT-5.6 Terra | Muse Spark 1.3 |
|---|---|---|
| Developer | OpenAI | Meta |
| Released | 9 Jul 2026 | 2 Sept 2026 |
| Context window (tokens) | 1,050,000 | 1,048,576 |
| Compared at | GPT-5.6 Terra (max) | Muse Spark 1.3 (max) |
| Open weights | No | No |
| Price in / out / cached input per 1M | $2.00 / $12.00 / $0.200 | $1.25 / $4.25 / $0.150 |
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-5.6 Terra (max) | Muse Spark 1.3 (max) | Leader |
|---|---|---|---|
| Reasoning | 70.1 | 77.4 | Muse Spark 1.3 |
| Coding | 65.1 | 64.5 | level |
| Agents & tools | 58.9 | 65.6 | Muse Spark 1.3 |
| Maths | 68.9 | 63.4 | GPT-5.6 Terra |
| Knowledge | 57.2 | 72.8 | Muse Spark 1.3 |
| Instruction following | 69.9 | – | not comparable |
| Human preference | – | 69.1 | not comparable |
| Multimodal | 63.7 | 65.7 | Muse Spark 1.3 |
| Long context | 66.6 | 66.6 | level |
| Composite | 76.6 | 83.3 | Muse Spark 1.3 |
Where sources disagree
Across the 37 benchmarks both have, GPT-5.6 Terra leads 16, Muse Spark 1.3 leads 19 and 2 are tied. Every overall index that rates both puts Muse Spark 1.3 ahead.
| Overall index | GPT-5.6 Terra (max) | Muse Spark 1.3 (max) | Leader |
|---|---|---|---|
| BenchLeader Index | 64.4 | 67.1 | neck and neck |
| AA Intelligence Index v4.3.2 | 42.1 | 48.1 | Muse Spark 1.3 |
| Arena Text | 1467† | 1496 | Muse Spark 1.3 |
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; suite cost is what Artificial Analysis paid to run its whole Intelligence Index.
| Measure | GPT-5.6 Terra (max) | Muse Spark 1.3 (max) |
|---|---|---|
| Blended price per 1M tokens | $4.50 | $2.00 |
| Input / output per 1M | $2.00 / $12.00 | $1.25 / $4.25 |
| Cached input per 1M | $0.200 | $0.150 |
| Price per task: chat reply | $0.0044 | $0.0018 |
| Price per task: summarize a 30-page report | $0.031 | $0.018 |
| Price per task: code edit | $0.030 | $0.014 |
| Price per task: agentic coding session | $0.168 | $0.092 |
| Price per task: structured extraction | $0.0064 | $0.0034 |
| Cost to run a full benchmark suite | $1.40 | $1.60 |
| Output speed | 126 tok/s | 116 tok/s |
| Time to first answer | 91.6 s | 90.1 s |
| Time to a complete answer | 95.6 s | 94.4 s |
Per token is not per task: on Artificial Analysis' full suite GPT-5.6 Terra (max) cost $1.40 and Muse Spark 1.3 (max) $1.60, because Muse Spark 1.3 (max) used about 2.6× as many tokens for the same work, which a per-token price hides.
Each model is compared at its best-ranked reasoning-effort setting. GPT-5.6 Terra (max) scores 64.4 and takes 92 s to its first answer token; GPT-5.6 Terra (xhigh) scores 62.8 in 7 s, GPT-5.6 Terra (high) scores 61.3 in 2 s. Muse Spark 1.3 (max) scores 67.1 and takes 90 s to its first answer token; Muse Spark 1.3 (xhigh) scores 64.9 in 30 s, Muse Spark 1.3 scores 59.8 in 3 s. 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
128 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 | GPT-5.6 Terra (max) | Muse Spark 1.3 (max) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 64.4 | 67.1 |
| Agents & tools score | BenchLeader | 58.9 | 65.6 |
| Coding score | BenchLeader | 65.1 | 64.5 |
| Composite score | BenchLeader | 76.6 | 83.3 |
| Instruction following score | BenchLeader | 69.9 | – |
| Knowledge score | BenchLeader | 57.2 | 72.8 |
| Long context score | BenchLeader | 66.6 | 66.6 |
| Maths score | BenchLeader | 68.9 | 63.4 |
| Multimodal score | BenchLeader | 63.7 | 65.7 |
| Reasoning score | BenchLeader | 70.1 | 77.4 |
| Human preference score | BenchLeader | – | 69.1 |
| Blended price $/M | BenchLeader | $4.50 | $2.00 |
| Output speed | BenchLeader | 126 tok/s | 116 tok/s |
| Time to first answer | BenchLeader | 91.6 s | 90.1 s |
| Context window | BenchLeader | 1,050,000 tokens | 1,048,576 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 93.3% | – |
| SimpleBench | SimpleBench | 48.9%† | – |
| Arena Hard Prompts | Arena (LMArena) | 1490† | 1517 |
| LiveBench 2.0 Reasoning | LiveBench | 90.6%† | 89.7%† |
| LiveBench 1.0 Reasoning | LiveBench | 90.6% | 89.7%† |
| GPQA Diamond (AA) | Artificial Analysis | 92.5% | 93.5% |
| Humanity's Last Exam (AA) | Artificial Analysis | 42.9% | 48.7% |
| GPQA Diamond (Vals) | Vals AI | 90.9%† | – |
| Kagi LLM Benchmark | Kagi LLM Benchmark | 51.3%† | – |
| ARC-AGI-1 | ARC Prize | 96.5% | – |
| ARC-AGI-2 | ARC Prize | 83.9% | – |
| ARC-AGI-3 | ARC Prize | 0.8% | – |
| CritPt | Artificial Analysis | 30.0% | 24.9% |
| MysteryMechanism | Vals AI | – | 36.0% |
| Chess Puzzles | Epoch AI Benchmarking Hub | 47.0% | 37.0% |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 35.0% | 25.0% |
| LMCA | Epoch AI Benchmarking Hub | 55.0% | 53.9%† |
| DTBench | Epoch AI Benchmarking Hub | 93.3% | 96.5%† |
| BTF-3 | Epoch AI Benchmarking Hub | – | 14.1%† |
| Coding | |||
| SciCode | SciCode | 55.0% | 58.8% |
| WeirdML | WeirdML | 78.3%† | – |
| FrontierCode | Cognition | 41.3%† | – |
| Arena Coding | Arena (LMArena) | 1518† | 1536 |
| Arena WebDev | Arena (LMArena) | 1521† | 1657 |
| LiveBench 2.0 Coding | LiveBench | 86.2%† | 90.2%† |
Show all 128 benchmarks
| Metric | Publisher | GPT-5.6 Terra (max) | Muse Spark 1.3 (max) |
|---|---|---|---|
| Coding | |||
| LiveBench 1.0 Coding | LiveBench | 78.3% | 81.1%† |
| LiveCodeBench | Vals AI | 85.9%† | – |
| IOI | Vals AI | 87.6% | 56.6% |
| SWE-bench (Vals) | Vals AI | 95.4% | – |
| Code Migration | Vals AI | 47.8% | 47.4% |
| ProgramBench | Vals AI | 0.5% | 2.5% |
| SWE-bench Multimodal | Vals AI | – | 43.1% |
| Vibe Code Bench 1-100 | Vals AI | 14.8% | 20.5% |
| Vibe Code Bench v1.1 | Vals AI | 74.6% | 85.9% |
| Surface Evolver Bench | Epoch AI Benchmarking Hub | 83.8%† | – |
| DeepSWE v1.1 | Epoch AI Benchmarking Hub | 69.6% | – |
| CursorBench | Cursor | 41.3% | 41.6% |
| ALE-Bench | Epoch AI Benchmarking Hub | 1951.4 | 1688.3† |
| FrontierSWE | Epoch AI Benchmarking Hub | – | 25.8% |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 35.4% | 33.3% |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | 88.0% | 84.3% |
| Startrise Frontend | Startrise | 66.9%† | 74.7%† |
| FrontierCode 1.1 (AA) | Artificial Analysis | – | 40.6% |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | 21.5% | 14.6%† |
| APEX-Agents | Mercor | 58.2% | 47.6% |
| Arena Agent | Arena (LMArena) | 0.5† | 3.4 |
| LiveBench Long Agentic | LiveBench | 48.4%† | 40.4%† |
| LiveBench Browser | LiveBench | 84.0%† | 80.0%† |
| LiveBench Agentic Coding | LiveBench | 55.0% | 64.1%† |
| Terminal-Bench Hard | Artificial Analysis | 57.6% | – |
| τ²-Bench Telecom (AA) | Artificial Analysis | 86.3% | – |
| Terminal-Bench 2.1 (Vals) | Vals AI | 77.5% | 79.0% |
| HiL-Bench | Scale AI SEAL | – | 66.9%† |
| GDPval-AA v2.1 | Artificial Analysis | 47.6% | – |
| τ³-Banking (AA) | Artificial Analysis | 40.2% | 50.5% |
| ITBench SRE (AA) | Artificial Analysis | 51.0% | 33.2% |
| Analyst Agent (AA) | Artificial Analysis | – | 37.5% |
| APEX-Agents (AA) | Artificial Analysis | 38.9% | – |
| CUA-bench | Vals AI | – | 5.8% |
| CyberBench | Vals AI | 72.1% | 72.7% |
| Finance Agent v2 | Vals AI | 54.4% | 60.0% |
| Harvey's Legal Agent Benchmark v1 | Vals AI | 0.8% | 23.8% |
| Legal Research Bench | Vals AI | 41.4% | 55.3% |
| SkillsBench | Vals AI | 58.9% | – |
| SREBench | Vals AI | 9.5% | 3.8% |
| Tax Agent Bench | Vals AI | 30.0% | 42.4% |
| Terminal-Bench 4.0 (Vals) | Vals AI | 22.7% | 24.8% |
| Terminal-Bench Science | Vals AI | 10.0% | 10.0% |
| BALROG | BALROG | 53.2% | – |
| Vending-Bench 2 | Andon Labs | 7343.2† | – |
| AutomationBench | Artificial Analysis | 59.6% | 57.9% |
| GDP.pdf (AA) | Artificial Analysis | 24.0% | 26.6% |
| Harvey LAB-AA v1.1 | Harvey | – | 8.9% |
| EnterpriseOps-Gym | Artificial Analysis | 38.5% | – |
| AA-Briefcase v1.1 | Artificial Analysis | – | 1581 |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | 13.8% | 10.9% |
| FACTS Search | 81.7%† | – | |
| Maths | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 86.0% | 74.0% |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 70.7% | 46.3% |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 99.7% | 99.2%† |
| ProofBench | Vals AI | 74.0%† | 58.0% |
| LiveBench Mathematics | LiveBench | 94.9% | 96.0%† |
| Arena Maths | Arena (LMArena) | 1483† | 1510 |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 43.2% | – |
| LiveBench 2.0 Data Analysis | LiveBench | 79.3%† | 79.6%† |
| LiveBench Research | LiveBench | 60.4%† | 68.6%† |
| LiveBench 1.0 Data Analysis | LiveBench | 79.3% | 79.6%† |
| AA-Omniscience | Artificial Analysis | 0.1 | 25 |
| MMLU-Pro | Vals AI | 86.7%† | – |
| LegalBench | Vals AI | 85.1% | – |
| CorpFin | Vals AI | 65.3%† | – |
| TaxEval | Vals AI | 76.2%† | – |
| PRBench Finance | Scale AI SEAL | – | 59.5%† |
| PRBench Legal | Scale AI SEAL | – | 61.6%† |
| Excel Modeling Benchmark | Vals AI | 66.2% | 67.4% |
| MedCode | Vals AI | 43.4%† | – |
| MedScribe | Vals AI | 82.9%† | – |
| Public Benefits Bench v1.1 | Vals AI | 62.4% | – |
| DrugDiscoveryBench | Scale AI SEAL | – | 62.2%† |
| GDP.pdf | Epoch AI Benchmarking Hub | 24.7%† | 27.6%† |
| AA-Omniscience: accuracy | Artificial Analysis | 46.8% | 43.6% |
| AA-Omniscience: non-hallucination | Artificial Analysis | 12.1% | 67.1% |
| Instruction following | |||
| LiveBench Language | LiveBench | 82.9% | 82.8%† |
| LiveBench Instruction Following | LiveBench | 64.6% | 78.0%† |
| IFBench | Artificial Analysis | 71.2% | – |
| Arena Instruction Following | Arena (LMArena) | 1464† | 1485 |
| Human preference | |||
| Arena Text | Arena (LMArena) | 1467† | 1496 |
| EQ-Bench 4 | EQ-Bench | 1234† | – |
| Arena Creative Writing | Arena (LMArena) | 1429† | 1463 |
| Arena Multi-turn | Arena (LMArena) | 1472† | 1494 |
| Arena Longer Queries | Arena (LMArena) | 1468† | 1497 |
| Multimodal | |||
| Arena Vision | Arena (LMArena) | 1271† | 1308 |
| LiveBench Design | LiveBench | 76.0%† | 74.8%† |
| MMMU-Pro | Artificial Analysis | 80.7% | 82.0%† |
| MMMU-Pro (Vals) | Vals AI | 86.5% | – |
| MortgageTax | Vals AI | 67.3%† | – |
| SAGE | Vals AI | 47.0% | – |
| Blueprint-Bench 2 | Epoch AI Benchmarking Hub | 30.8%† | – |
| AA-3D | Artificial Analysis | – | 1037 |
| Long context | |||
| AA-LCR | Artificial Analysis | 83.0% | 83.0% |
| Arena Document | Arena (LMArena) | 1477† | 1468 |
| MLCR | Artificial Analysis | 27.2% | 13.9% |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 160.1† | 157.2† |
| LiveBench 2.0 | LiveBench | 75.0%† | 74.8%† |
| LiveBench 1.0 | LiveBench | 77.9% | 81.6%† |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 42.1 | 48.1 |
| Vals RSI Index | Vals AI | – | 19.6% |
| Vals Multimodal Index | Vals AI | 65.1%† | – |
Data as of 2026-10-24. Best configuration of each model; every score links to its source on the model pages.
What changed
- GPT-5.6 Terra (max): 1 new result (Harvey's Legal Agent Benchmark v1).
- Muse Spark 1.3 (max): 3 new results (Harvey's Legal Agent Benchmark v1, Vals RSI Index, SWE-bench Multimodal); 1 score revision.
- GPT-5.6 Terra (max): 1 score revision.
- Muse Spark 1.3 (max): 1 score revision.
- Muse Spark 1.3 (max): 1 score revision.
- GPT-5.6 Terra (max): BenchLeader Index 63.8 → 64.4.
- GPT-5.6 Terra (max): 5 new results (AutomationBench, GDP.pdf (AA), MLCR…).
- GPT-5.6 Terra (max): 1 new result (SREBench); 1 score revision.
- Muse Spark 1.3 (max): 2 new results (ProgramBench, SREBench); 1 score revision.
- GPT-5.6 Terra (max): 1 score revision.
Related comparisons
- GPT-5.6 Terra vs Claude Opus 5.5
- GPT-5.6 Terra vs Gemini 4 Argon
- GPT-5.6 Terra vs Muse Wonder Full
- Muse Spark 1.3 vs Claude Opus 5.5
- Muse Spark 1.3 vs Gemini 4 Argon
- Muse Spark 1.3 vs Muse Wonder Full
- Muse Spark 1.3 vs Claude Fable 5.1
- GPT-5.5 vs Muse Spark 1.3
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
GPT-5.6 Terra vs Muse Spark 1.3: questions
- Is GPT-5.6 Terra better than Muse Spark 1.3?
- GPT-5.6 Terra (max) and Muse Spark 1.3 (max) are neck and neck on quality (64.4 vs 67.1, within the ±6.7 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 2026-10-24, so check the category scores for your use.
- Which is cheaper, GPT-5.6 Terra or Muse Spark 1.3, and will that change?
- Muse Spark 1.3 is cheaper today: $2.00 against $4.50 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is GPT-5.6 Terra better than Muse Spark 1.3 for coding?
- They are level in coding (65 vs 65 on the category index). On SciCode, the top shared coding benchmark, it is 55.0% for GPT-5.6 Terra against 58.8% for Muse Spark 1.3.
- Is GPT-5.6 Terra better than Muse Spark 1.3 for agents and tool use?
- Muse Spark 1.3 scores higher in agentic tasks (66 vs 59 on the category index, where 50 is average). On τ³-Banking (AA), the top shared agents & tools benchmark, it is 40.2% for GPT-5.6 Terra against 50.5% for Muse Spark 1.3.
- Which is better for long documents?
- They are level on long-context benchmarks (67 vs 67). Both take about 1.1M tokens: GPT-5.6 Terra 1,050,000, Muse Spark 1.3 1,048,576.
- Which has lower latency?
- Muse Spark 1.3 (max) answers first: 90.1 s against 91.6 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
- Which is faster, GPT-5.6 Terra or Muse Spark 1.3?
- GPT-5.6 Terra streams faster: 126 against 116 output tokens per second.
- Do independent benchmarks agree on GPT-5.6 Terra vs Muse Spark 1.3?
- Across the 37 benchmarks both have, GPT-5.6 Terra leads 16, Muse Spark 1.3 leads 19 and 2 are tied. Every overall index that rates both puts Muse Spark 1.3 ahead.
- When was Muse Spark 1.3 released?
- Muse Spark 1.3 was released on 2 Sept 2026.