OpenAIGPT-5.6 Luna
Reasoning effort
GPT-5.6 Luna is an OpenAI proprietary model, released 9 Jul 2026. Its best configuration (max reasoning effort) ranks #120 of 785 on the BenchLeader Index at 59.9 ±3.9, in the upper half. The ± is the point: 220 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in composite (71) and lowest in agents & tools (49). At $0.450 per million tokens blended it is cheaper than most ranked models. Output speed of 129 tokens per second puts it faster than most, with a first answer in 62.0 s. It has been measured at 7 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 20 Oct 2026.
- Blended price
- $0.450/M
- $0.200 in · $1.20 out
- Output speed
- 129 tok/s
- measured by Artificial Analysis
- First answer
- 62 s
- first token 1.78 s
- Context
- 1.1M
- Full answer
- 66 s
- median, reasoning included
- Cost per run
- $0.178
- one full Intelligence Index run
- Released
- 9 Jul 2026
- Overall index60
- Reasoning63
- Coding56
- Agents & tools49
- Maths65
- Knowledge57
- Multimodal62
- Long context67
- Composite71
Versions
OpenAI has shipped 2 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.
| Model | Released | Index | Rank |
|---|---|---|---|
| GPT-6 Lunamax | 22 Sept 2026 | 60.0 | #117 |
| GPT-5.6 Lunamaxthis page | 9 Jul 2026 | 59.9 | #120 |
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | 44.6 | #531 | 129 tok/s | 0.74 s | $0.0004 | Agents & tools 37 · Coding 41 · Composite 47 · Knowledge 50 · Long context 46 · Maths 45 · Multimodal 43 · Reasoning 38 |
| low | 50.8 | #358 | 121 tok/s | 1.44 s | $0.0004 | Agents & tools 41 · Coding 49 · Composite 53 · Knowledge 55 · Long context 60 · Maths 50 · Multimodal 57 · Reasoning 45 |
| medium | 52.5 | #315 | 129 tok/s | 1.79 s | $0.0004 | Agents & tools 46 · Coding 50 · Composite 58 · Knowledge 56 · Long context 62 · Multimodal 59 · Reasoning 43 |
| high | 57.0 | #189 | 126 tok/s | 7.61 s | $0.0004 | Agents & tools 54 · Coding 57 · Composite 65 · Knowledge 56 · Long context 65 · Multimodal 61 · Reasoning 55 |
| xhigh | 59.4 | #130 | 125 tok/s | 28 s | $0.0004 | Agents & tools 53 · Coding 57 · Composite 68 · Human preference 64 · Knowledge 57 · Long context 66 · Maths 66 · Multimodal 60 · Reasoning 58 |
| maxbest | 59.9 | #120 | 129 tok/s | 62 s | $0.0004 | Agents & tools 49 · Coding 56 · Composite 71 · Knowledge 57 · Long context 67 · Maths 65 · Multimodal 62 · Reasoning 63 |
| not stated | 48.0 | #442 | 68 tok/s | 1.78 s | $0.0004 | Agents & tools 43 · Coding 53 · Composite 48 · Human preference 44 · Reasoning 48 |
What more thinking costs
Turning the effort up buys index points and multiplies the bill. Here max costs 18× low for +9.1 on the index — 2.2 points per doubling of spend. Compare that with other models.
How its index has moved
8 Sept 2026 to 23 Oct 2026- max
- xhigh
- high
- medium
The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | no reasoning | low | medium | high | xhigh | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| GPQA Diamond | 63.6%#188 | 82.3%#113 | – | – | – | 91.6%#31 | – | Epoch AI Benchmarking Hub | |
| SimpleBench | – | – | – | – | 46.8%#48 | – | 46.8%#48 | SimpleBench | |
| Arena Hard Prompts | – | – | – | – | 1475#75 | – | – | Arena (LMArena) | |
| LiveBench 2.0 Reasoningnot in indexeffort not stated | – | – | – | – | – | – | 85.6%#30 | LiveBench | |
| LiveBench 1.0 Reasoningnot in index | – | – | – | – | – | 85.6%#36 | – | LiveBench | |
| GPQA Diamond (AA)not in index | 64.5%#346 | 83.5%#156 | 85.9%#117 | 89.2%#80 | 89.5%#74 | 91.1%#50 | – | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 7.2%#371 | 19.8%#216 | 25.8%#180 | 33.4%#136 | 37.0%#109 | 39.5%#90 | – | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | – | – | – | – | – | 91.7%#18 | – | Vals AI | |
| Kagi LLM Benchmarkeffort not stated | – | – | – | – | – | – | 49.1%#86 | Kagi LLM Benchmark | |
| ARC-AGI-1 | 3.3%#252 | 39.5%#194 | 56.5%#170 | 78.0%#124 | 87.7%#92 | 90.7%#73 | – | ARC Prize | |
| ARC-AGI-2 | 0.8%#231 | 5.1%#176 | 7.4%#167 | 33.3%#124 | 47.6%#110 | 59.6%#90 | – | ARC Prize | |
| ARC-AGI-3 | – | 0.2%#58 | 0.2%#58 | 0.1%#64 | 0.0%#67 | 0.2%#58 | – | ARC Prize | |
| CritPt | 0.3%#271 | 2.6%#189 | 4.9%#158 | 16.6%#85 | 20.6%#61 | 20.6%#61 | – | Artificial Analysis | |
| MysteryMechanismnot in index | – | – | – | – | – | 14.4%#27 | – | Vals AI | |
| Chess Puzzlesnot in index | 3.0%#135 | 17.0%#102 | – | – | – | 43.0%#34 | – | Epoch AI Benchmarking Hub | |
| Mystery Game Puzzlesnot in index | 20.0%#60 | 17.0%#75 | 12.0%#103 | – | – | 21.0%#58 | – | Epoch AI Benchmarking Hub | |
| LMCAnot in index | 41.4%#103 | 43.8%#99 | 44.3%#95 | 47.4%#79 | 48.5%#74 | 48.5%#74 | – | Epoch AI Benchmarking Hub | |
| DTBenchnot in index | 69.3%#158 | 80.0%#132 | 83.2%#114 | 86.7%#104 | 89.1%#92 | 88.8%#94 | – | Epoch AI Benchmarking Hub |
Coding
| Benchmark | no reasoning | low | medium | high | xhigh | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| SciCode | 40.4%#158 | 46.1%#135 | 46.8%#130 | 51.6%#93 | 50.5%#103 | 53.6%#72 | – | SciCode | |
| WeirdML | – | – | – | 60.9%#46 | – | – | – | WeirdML | |
| FrontierCodeeffort not stated | – | – | – | – | – | – | 39.8%#27 | Cognition | |
| Arena Coding | – | – | – | – | 1502#75 | – | – | Arena (LMArena) | |
| Arena WebDev | – | – | – | – | 1519#54 | – | – | Arena (LMArena) | |
| LiveBench 2.0 Codingnot in indexeffort not stated | – | – | – | – | – | – | 73.6%#37 | LiveBench | |
| LiveBench 1.0 Codingnot in index | – | – | – | – | – | 82.9%#9 | – | LiveBench | |
| SciCode (AA)not in index | 40.4%#170 | 46.1%#140 | 46.8%#137 | 51.6%#93 | 50.5%#105 | 53.6%#76 | – | Artificial Analysis | |
| IOI | – | – | – | – | – | 61.8%#16 | – | Vals AI | |
| SWE-bench (Vals)not in index | – | – | – | – | – | 93.0%#9 | – | Vals AI | |
| Code Migrationnot in index | – | – | – | – | – | 44.5%#21 | – | Vals AI | |
| ProgramBenchnot in index | – | – | – | – | – | 0.0%#32 | – | Vals AI | |
| Vibe Code Bench 1-100not in index | – | – | – | – | – | 22.6%#8 | – | Vals AI | |
| Vibe Code Bench v1.1not in index | – | – | – | – | – | 77.1%#33 | – | Vals AI | |
| Surface Evolver Benchnot in index | – | – | 61.9%#12 | – | – | – | – | Epoch AI Benchmarking Hub | |
| DeepSWE v1.1not in index | – | 1.6%#69 | 11.3%#68 | 44.3%#53 | 56.9%#37 | 67.2%#21 | – | Epoch AI Benchmarking Hub | |
| CursorBenchnot in index | – | 16.0%#78 | 22.2%#77 | 29.4%#67 | 33.0%#59 | 35.9%#49 | – | Cursor | |
| ALE-Benchnot in index | – | – | – | – | – | 1667.4#15 | – | Epoch AI Benchmarking Hub | |
| Terminal-Bench 4.0 (AA)not in index | 1.0%#128 | 0.0%#152 | 0.5%#139 | 2.5%#113 | 3.5%#107 | 11.6%#86 | – | Artificial Analysis | |
| Terminal-Bench 2.1 (AA)not in index | 39.0%#141 | 43.5%#133 | 53.2%#112 | 69.7%#73 | 77.9%#54 | 80.9%#41 | – | Artificial Analysis | |
| Startrise Frontendnot in indexeffort not stated | – | – | – | – | – | – | 65.0%#25 | Startrise |
Agents & tools
| Benchmark | no reasoning | low | medium | high | xhigh | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Terminal-Bench | – | – | – | – | – | 17.3%#74 | – | Terminal-Bench | |
| APEX-Agentseffort not stated | – | – | – | – | – | – | 43.0%#40 | Mercor | |
| Arena Agent | – | – | – | – | -2.3#34 | – | – | Arena (LMArena) | |
| LiveBench Long Agenticnot in indexeffort not stated | – | – | – | – | – | – | 30.4%#32 | LiveBench | |
| LiveBench Browsernot in indexeffort not stated | – | – | – | – | – | – | 68.0%#24 | LiveBench | |
| LiveBench Agentic Codingnot in index | – | – | – | – | – | 48.4%#48 | – | LiveBench | |
| Terminal-Bench 2.1 (Vals) | – | – | – | – | – | 79.0%#11 | – | Vals AI | |
| τ³-Banking (AA) | 9.7%#146 | 12.8%#130 | 17.7%#105 | 25.1%#82 | 28.7%#72 | 31.1%#63 | – | Artificial Analysis | |
| ITBench SRE (AA)not in index | – | – | – | – | – | 40.3%#18 | – | Artificial Analysis | |
| APEX-Agents (AA)not in index | – | – | – | – | – | 35.8%#7 | – | Artificial Analysis | |
| CyberBenchnot in index | – | – | – | – | – | 73.6%#11 | – | Vals AI | |
| Finance Agent v2not in index | – | – | – | – | – | 55.0%#24 | – | Vals AI | |
| Harvey's Legal Agent Benchmark v1not in index | – | – | – | – | – | 1.3%#52 | – | Vals AI | |
| Legal Research Benchnot in index | – | – | – | – | – | 36.5%#41 | – | Vals AI | |
| SkillsBenchnot in index | – | – | – | – | – | 60.5%#7 | – | Vals AI | |
| SREBenchnot in index | – | – | – | – | – | 1.5%#25 | – | Vals AI | |
| Tax Agent Benchnot in index | – | – | – | – | – | 24.8%#39 | – | Vals AI | |
| Terminal-Bench 4.0 (Vals)not in index | – | – | – | – | – | 11.6%#37 | – | Vals AI | |
| Terminal-Bench Sciencenot in index | – | – | – | – | – | 0.0%#38 | – | Vals AI | |
| BALROGnot in index | – | – | – | – | – | 45.6%#8 | – | BALROG | |
| Vending-Bench 2not in indexeffort not stated | – | – | – | – | – | – | 4094.7#41 | Andon Labs | |
| AutomationBenchnot in index | 5.0%#165 | 11.7%#153 | 23.1%#129 | 35.6%#107 | 42.6%#93 | 50.2%#78 | – | Artificial Analysis | |
| GDP.pdf (AA)not in index | 6.0%#154 | 14.2%#102 | 15.2%#97 | 22.2%#50 | 23.8%#41 | 24.0%#38 | – | Artificial Analysis | |
| EnterpriseOps-Gymnot in index | – | – | – | – | – | 40.8%#25 | – | Artificial Analysis | |
| Terminal-Bench-Science 0.1 (AA)not in index | – | – | – | – | – | 3.3%#55 | – | Artificial Analysis | |
| FACTS Searchnot in indexeffort not stated | – | – | – | – | – | – | 79.3%#15 |
Maths
| Benchmark | no reasoning | low | xhigh | max | Source | Trend |
|---|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | 39.6%#62 | 41.4%#61 | – | 82.1%#15 | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | – | – | – | 61.0%#22 | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | 40.0%#198 | 66.7%#149 | – | 98.3%#24 | Epoch AI Benchmarking Hub | |
| ProofBench | – | – | – | 60.0%#20 | Vals AI | |
| LiveBench Mathematicsnot in index | – | – | – | 87.2%#46 | LiveBench | |
| Arena Maths | – | – | 1478#40 | – | Arena (LMArena) |
Knowledge
| Benchmark | no reasoning | low | medium | high | xhigh | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | – | – | 41.0%#46 | – | Epoch AI Benchmarking Hub | |
| LiveBench 2.0 Data Analysisnot in indexeffort not stated | – | – | – | – | – | – | 78.0%#20 | LiveBench | |
| LiveBench Researchnot in indexeffort not stated | – | – | – | – | – | – | 51.0%#46 | LiveBench | |
| LiveBench 1.0 Data Analysisnot in index | – | – | – | – | – | 78.0%#28 | – | LiveBench | |
| AA-Omniscience | -24.9#261 | -14.7#220 | -13.2#215 | -12#209 | -10.8#199 | -10.3#194 | – | Artificial Analysis | |
| MMLU-Pro | – | – | – | – | – | 86.0%#48 | – | Vals AI | |
| LegalBench | – | – | – | – | – | 84.0%#43 | – | Vals AI | |
| CorpFin | – | – | – | – | – | 64.2%#39 | – | Vals AI | |
| TaxEval | – | – | – | – | – | 76.2%#6 | – | Vals AI | |
| BioMysteryBenchnot in index | – | – | – | – | – | 61.5%#23 | – | Vals AI | |
| Excel Modeling Benchmarknot in index | – | – | – | – | – | 67.1%#18 | – | Vals AI | |
| MedCodenot in index | – | – | – | – | – | 42.4%#50 | – | Vals AI | |
| MedScribenot in index | – | – | – | – | – | 84.4%#35 | – | Vals AI | |
| Public Benefits Bench v1.1not in index | – | – | – | – | – | 61.2%#30 | – | Vals AI | |
| GDP.pdfnot in index | – | – | 22.7%#27 | – | – | – | 22.7%#27 | Epoch AI Benchmarking Hub | |
| AA-Omniscience: accuracynot in index | 28.6%#195 | 39.6%#126 | 40.7%#120 | 41.8%#110 | 42.5%#108 | 42.7%#107 | – | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 25.0%#243 | 10.1%#418 | 9.1%#435 | 7.6%#479 | 7.5%#481 | 7.4%#483 | – | Artificial Analysis |
Instruction following
| Benchmark | xhigh | max | Source | Trend |
|---|---|---|---|---|
| LiveBench Languagenot in index | – | 72.6%#56 | LiveBench | |
| LiveBench Instruction Followingnot in index | – | 60.1%#57 | LiveBench | |
| Arena Instruction Followingnot in index | 1444#73 | – | Arena (LMArena) |
Human preference
| Benchmark | xhigh | not stated | Source | Trend |
|---|---|---|---|---|
| Arena Text | 1453#73 | – | Arena (LMArena) | |
| EQ-Bench 4effort not stated | – | 1156#18 | EQ-Bench | |
| Arena Creative Writingnot in index | 1410#90 | – | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1459#77 | – | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1452#85 | – | Arena (LMArena) |
Multimodal
| Benchmark | no reasoning | low | medium | high | xhigh | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Arena Vision | – | – | – | – | 1257#58 | – | – | Arena (LMArena) | |
| LiveBench Designnot in indexeffort not stated | – | – | – | – | – | – | 76.0%#26 | LiveBench | |
| MMMU-Pro | 60.4%#210 | 74.0%#118 | 75.8%#95 | 77.6%#81 | 78.5%#69 | 78.5%#69 | – | Artificial Analysis | |
| MMMU-Pro (Vals)not in index | – | – | – | – | – | 85.0%#25 | – | Vals AI | |
| MortgageTaxnot in index | – | – | – | – | – | 67.3%#27 | – | Vals AI | |
| SAGEnot in index | – | – | – | – | – | 44.2%#46 | – | Vals AI | |
| Blueprint-Bench 2not in indexeffort not stated | – | – | – | – | – | – | 22.6%#25 | Epoch AI Benchmarking Hub |
Long context
| Benchmark | no reasoning | low | medium | high | xhigh | max | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| AA-LCR | 42.7%#348 | 70.0%#215 | 75.0%#157 | 80.3%#71 | 81.7%#52 | 83.7%#16 | Artificial Analysis | |
| Arena Documentnot in index | – | – | – | – | 1462#23 | – | Arena (LMArena) | |
| MLCRnot in index | 8.9%#122 | 15.6%#72 | 13.3%#90 | 16.7%#67 | 15.0%#78 | 20.0%#47 | Artificial Analysis |
Composite
| Benchmark | no reasoning | low | medium | high | xhigh | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in indexeffort not stated | – | – | – | – | – | – | 156.5#29 | Epoch AI Benchmarking Hub | |
| LiveBench 2.0effort not stated | – | – | – | – | – | – | 66.1%#31 | LiveBench | |
| LiveBench 1.0not in index | – | – | – | – | – | 73.6%#45 | – | LiveBench | |
| AA Intelligence Index v4.3.2 | 15.5#281 | 21.0#218 | 25.0#166 | 32.1#112 | 34.6#91 | 37.3#83 | – | Artificial Analysis | |
| Vals Science Indexnot in index | – | – | – | – | – | 25.3%#21 | – | Vals AI | |
| Vals Multimodal Indexnot in index | – | – | – | – | – | 69.1%#7 | – | Vals AI |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Azure | 135 tok/s | 5.66 s | $0.220 | $1.32 | 1.1M | – |
| OpenAI | 114 tok/s | 1.79 s | $0.400 | $2.40 | 1.1M | – |
| Azure | 94 tok/s | 2.96 s | $0.200 | $1.20 | 1.1M | – |
| Amazon Bedrock | 92 tok/s | 0.85 s | $0.220 | $1.32 | 1.1M | – |
| Azure | 91 tok/s | 1.46 s | $0.220 | $1.32 | 1.1M | – |
| OpenAI | 66 tok/s | 1.49 s | $0.200 | $1.20 | 1.1M | – |
| OpenAI | 60 tok/s | 1.69 s | $0.100 | $0.600 | 1.1M | – |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.020 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0004 | $0.0004 | 1.1 min |
| Summarize a 30-page report | 12,000 / 600 | $0.0031 | $0.0015 | 1.1 min |
| Code edit | 6,000 / 1,500 | $0.0030 | $0.0022 | 1.2 min |
| Agentic coding session | 60,000 / 4,000 | $0.017 | $0.0087 | 1.5 min |
| Structured extraction | 2,000 / 200 | $0.0006 | $0.0004 | 1.1 min |
More from OpenAI
Compare GPT-5.6 Luna with…
See also
Questions people ask
- What are GPT-5.6 Luna's benchmark scores?
- As of 24 Oct 2026, GPT-5.6 Luna (max effort) scores 59.9 ±3.9 on the BenchLeader Index, strongest in composite (71) and weakest in agents & tools (49). That rests on 23 independent benchmark results, and every score on this page links to its publisher.
- How does GPT-5.6 Luna rank?
- GPT-5.6 Luna (max effort) is #120 of 785 ranked configurations on 24 Oct 2026. 220 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does GPT-5.6 Luna cost per 1M tokens?
- As of 24 Oct 2026, GPT-5.6 Luna costs $0.200 per 1M input tokens and $1.20 per 1M output tokens, $0.450 blended (three input tokens per output token).
- Is GPT-5.6 Luna better than GPT-5.6 Sol?
- As of 24 Oct 2026, GPT-5.6 Sol (max) leads on quality (68.0 vs 59.9 on the BenchLeader Index), but GPT-5.6 Luna (max) costs 18× less, so it is the better buy unless you need the extra quality. The head-to-head page has the benchmark-by-benchmark detail.
- When was GPT-5.6 Luna released?
- OpenAI released GPT-5.6 Luna on 9 Jul 2026. BenchLeader has results for it at 7 reasoning-effort settings.
Data as of 24 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “GPT-5.6 Luna: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/gpt-5-6-luna.