OpenAIGPT-5.6 Luna
Best configuration ranks #105 of 610 on the BenchLeader Index at 59.3 ±4.0 (xhigh reasoning effort). Last measured 8 Sept 2026. Released 9 Jul 2026.
- Blended price
- $0.450/M
- $0.200 in · $1.20 out
- Output speed
- 106 tok/s
- First answer
- 60 s
- first token 3.34 s
- Context
- 1M
- Overall index59
- Reasoning54
- Coding60
- Agents & tools50
- Knowledge59
- Human preference65
- Multimodal62
- Long context67
- Composite73
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. “Default” means the publisher did not say which setting was used.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | 46.9 | #362 | 116 tok/s | 0.80 s | $0.0004 | Coding 44 · Composite 50 · Knowledge 52 · Long context 47 · Maths 47 · Multimodal 45 · Reasoning 37 |
| low | 53.6 | #218 | 109 tok/s | 1.77 s | $0.0004 | Coding 52 · Composite 57 · Knowledge 57 · Long context 61 · Maths 52 · Multimodal 58 · Reasoning 46 |
| medium | 54.4 | #199 | 109 tok/s | 2.77 s | $0.0004 | Coding 52 · Composite 62 · Knowledge 58 · Long context 64 · Multimodal 60 · Reasoning 41 |
| high | 58.1 | #128 | 113 tok/s | 17 s | $0.0004 | Coding 58 · Composite 70 · Knowledge 58 · Long context 67 · Multimodal 62 · Reasoning 49 |
| xhighbest | 59.3 | #105 | 106 tok/s | 60 s | $0.0004 | Agents & tools 50 · Coding 60 · Composite 73 · Human preference 65 · Knowledge 59 · Long context 67 · Multimodal 62 · Reasoning 54 |
| max | 56.5 | #157 | 112 tok/s | 178 s | $0.0005 | Agents & tools 43 · Coding 71 · Composite 47 · Knowledge 57 · Maths 68 · Reasoning 60 |
| default | 56.6 | #155 | 112 tok/s | 178 s | $0.0004 | Coding 56 · Composite 76 · Human preference 44 · Knowledge 59 · Long context 69 · Multimodal 63 · Reasoning 47 |
Benchmark results
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 | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| GPQA Diamond | 63.6%#174 | 82.3%#101 | – | – | – | 91.6%#25 | – | Epoch AI Benchmarking Hub | |
| SimpleBench | – | – | – | – | 46.8%#48 | – | 46.8%#48 | SimpleBench | |
| LMArena Hard Prompts | – | – | – | – | 1473#67 | – | – | LMArena | |
| LiveBench Reasoningnot in index | – | – | – | – | – | 85.6%#27 | – | LiveBench | |
| GPQA Diamond (AA)not in index | 64.5%#330 | 83.5%#154 | 85.9%#116 | 89.2%#80 | 89.5%#74 | – | 91.1%#50 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 7.2%#300 | 19.8%#160 | 25.8%#131 | 33.4%#91 | 37.0%#73 | – | 39.5%#58 | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | – | – | – | – | – | 91.7%#18 | – | Vals AI | |
| Kagi LLM Benchmark | – | – | – | – | – | – | 49.1%#86 | Kagi LLM Benchmark | |
| ARC-AGI-1 | 3.3%#183 | 39.5%#134 | 56.5%#115 | 78.0%#85 | 87.7%#58 | 90.7%#49 | – | ARC Prize | |
| ARC-AGI-2 | 0.8%#164 | 5.1%#119 | 7.4%#112 | 33.3%#84 | 47.6%#74 | 59.6%#60 | – | ARC Prize | |
| ARC-AGI-3 | – | 0.2%#30 | 0.2%#30 | 0.1%#35 | 0.0%#38 | 0.2%#30 | – | ARC Prize |
Coding
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| SciCode | 39.9%#109 | 45.6%#83 | 45.8%#82 | 50.7%#54 | 50.0%#60 | 52.5%#48 | – | SciCode | |
| WeirdML | – | – | – | 60.9%#43 | – | – | – | WeirdML | |
| FrontierCode | – | – | – | – | – | – | 39.8%#15 | Cognition | |
| LMArena Coding | – | – | – | – | 1497#68 | – | – | LMArena | |
| LMArena WebDev | – | – | – | – | 1520#34 | – | – | LMArena | |
| LiveBench Codingnot in index | – | – | – | – | – | 82.9%#5 | – | LiveBench | |
| SciCode (AA)not in index | – | 46.1%#87 | 46.8%#84 | 51.6%#56 | 50.5%#65 | – | 53.6%#46 | Artificial Analysis | |
| IOI | – | – | – | – | – | 72.9%#4 | – | Vals AI | |
| SWE-bench (Vals)not in index | – | – | – | – | – | 93.0%#9 | – | Vals AI |
Agents & tools
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Terminal-Bench | – | – | – | – | – | 17.3%#58 | – | Terminal-Bench | |
| LMArena Agent | – | – | – | – | 1.1#23 | – | – | LMArena | |
| LiveBench Agentic Codingnot in index | – | – | – | – | – | 48.4%#35 | – | LiveBench | |
| Terminal-Bench 2.1 (Vals) | – | – | – | – | – | 79.0%#8 | – | Vals AI |
Maths
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | 39.6%#47 | 41.4%#46 | – | – | – | 82.1%#10 | – | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | – | – | – | – | – | 61.0%#16 | – | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | 40.0%#180 | 66.7%#134 | – | – | – | 98.3%#14 | – | Epoch AI Benchmarking Hub | |
| ProofBench | – | – | – | – | – | 60.0%#10 | – | Vals AI | |
| LiveBench Mathematicsnot in index | – | – | – | – | – | 87.2%#33 | – | LiveBench |
Knowledge
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | – | – | 41.0%#36 | – | Epoch AI Benchmarking Hub | |
| LiveBench Data Analysisnot in index | – | – | – | – | – | 78.0%#20 | – | LiveBench | |
| AA-Omniscience | -24.9#198 | -14.7#161 | -13.2#157 | -12#153 | -10.8#145 | – | -10.3#141 | Artificial Analysis | |
| MMLU-Pro | – | – | – | – | – | 86.0%#48 | – | Vals AI | |
| LegalBench | – | – | – | – | – | 84.0%#39 | – | Vals AI | |
| CorpFin | – | – | – | – | – | 64.2%#39 | – | Vals AI | |
| TaxEval | – | – | – | – | – | 76.2%#6 | – | Vals AI |
Instruction following
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| LiveBench Languagenot in index | – | – | – | – | – | 72.6%#45 | – | LiveBench |
Human preference
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| LMArena Text | – | – | – | – | 1453#63 | – | – | LMArena | |
| EQ-Bench 4 | – | – | – | – | – | – | 1156#18 | EQ-Bench |
Multimodal
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| LMArena Vision | – | – | – | – | 1254#45 | – | – | LMArena | |
| MMMU-Pro | 60.4%#176 | 74.0%#90 | 75.8%#67 | 77.6%#58 | 78.5%#48 | – | 78.5%#48 | Artificial Analysis |
Long context
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| AA-LCR | 42.7%#279 | 70.0%#161 | 75.0%#111 | 80.3%#45 | 81.7%#30 | – | 83.7%#8 | Artificial Analysis |
Composite
| Benchmark | no reasoning | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | – | – | – | – | – | 156.3#18 | Epoch AI Benchmarking Hub | |
| LiveBench | – | – | – | – | – | 73.6%#35 | – | LiveBench | |
| AA Intelligence Index | 16.8#205 | 21.8#155 | 25.8#112 | 32.9#68 | 34.8#54 | – | 37.5#47 | Artificial Analysis | |
| Vals Indexnot in index | – | – | – | – | – | 59.9#10 | – | 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 (EU) | 90 tok/s | 2.75 s | $0.220 | $1.32 | 1.1M | – |
| Amazon Bedrock (US) | 86 tok/s | 2.25 s | $0.220 | $1.32 | 1.1M | – |
| OpenAI | 73 tok/s | 2.36 s | $0.200 | $1.20 | 1.1M | – |
| OpenAI Fast | 63 tok/s | 3.34 s | $0.400 | $2.40 | 1.1M | – |
| Azure | 39 tok/s | 4.30 s | $0.200 | $1.20 | 1.1M | – |
| OpenAI Flex | 35 tok/s | 3.80 s | $0.100 | $0.600 | 1.1M | – |
| Azure (US) | 33 tok/s | 4.53 s | $0.220 | $1.32 | 1.1M | – |
Price history
Listed price per 1M tokens over time, as recorded by OpenRouter for the provider with the longest history.
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 $0.022 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 |
| Summarise 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 | $0.0088 | 1.6 min |
| Structured extraction | 2,000 / 200 | $0.0006 | $0.0004 | 1.0 min |
See also
Data as of 9 Sept 2026. Compare these configurations.