OpenAIGPT-5.5
Best configuration ranks #17 of 610 on the BenchLeader Index at 66.6 ±3.9. Last measured 8 Sept 2026. Released 23 Apr 2026.
- Blended price
- $11.25/M
- $5.00 in · $30.00 out
- Output speed
- 88 tok/s
- First answer
- 62 s
- first token 3.50 s
- Context
- 1.1M
- Overall index67
- Reasoning70
- Coding57
- Agents & tools68
- Knowledge74
- Instruction following73
- Human preference68
- Multimodal65
- Long context69
- Composite78
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 | 56.2 | #160 | 79 tok/s | 0.99 s | $0.011 | Agents & tools 76 · Coding 58 · Composite 58 · Instruction following 48 · Knowledge 62 · Long context 58 · Maths 49 · Multimodal 56 · Reasoning 57 |
| low | 62.3 | #57 | 75 tok/s | 1.65 s | $0.011 | Agents & tools 79 · Coding 60 · Composite 68 · Instruction following 63 · Knowledge 71 · Long context 67 · Maths 61 · Multimodal 64 · Reasoning 57 |
| medium | 64.9 | #30 | 82 tok/s | 7.61 s | $0.011 | Agents & tools 84 · Coding 62 · Composite 72 · Instruction following 69 · Knowledge 73 · Long context 68 · Multimodal 66 · Reasoning 66 |
| high | 66.3 | #19 | 78 tok/s | 14 s | $0.011 | Agents & tools 75 · Coding 63 · Composite 76 · Human preference 68 · Instruction following 70 · Knowledge 73 · Long context 69 · Multimodal 65 · Reasoning 64 |
| xhigh | 63.7 | #43 | 88 tok/s | 62 s | $0.012 | Agents & tools 56 · Coding 65 · Composite 67 · Knowledge 64 · Maths 71 · Reasoning 70 |
| defaultbest | 66.6 | #17 | 88 tok/s | 62 s | $0.011 | Agents & tools 68 · Coding 57 · Composite 78 · Human preference 68 · Instruction following 73 · Knowledge 74 · Long context 69 · Multimodal 65 · Reasoning 70 |
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 | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| GPQA Diamond | 77.3%#124 | 90.7%#34 | – | – | – | – | Epoch AI Benchmarking Hub | |
| SimpleBench | – | – | – | – | – | 69.0%#15 | SimpleBench | |
| LMArena Hard Prompts | – | – | – | 1502#23 | – | 1498#26 | LMArena | |
| LiveBench Reasoningnot in index | – | – | – | – | 89.7%#9 | – | LiveBench | |
| GPQA Diamond (AA)not in index | 76.8%#220 | 91.0%#54 | 92.6%#32 | 93.2%#22 | – | 93.5%#14 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 13.7%#207 | 32.7%#95 | 42.4%#40 | 45.0%#29 | – | 45.8%#27 | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | – | – | – | – | 93.2%#10 | – | Vals AI | |
| Kagi LLM Benchmark | – | – | – | – | – | 88.8%#2 | Kagi LLM Benchmark | |
| ARC-AGI-1 | – | 76.2%#90 | 92.2%#39 | 94.5%#24 | 95.0%#22 | – | ARC Prize | |
| ARC-AGI-2 | – | 33.3%#84 | 70.4%#37 | 83.3%#24 | 85.0%#18 | – | ARC Prize | |
| ARC-AGI-3 | – | – | – | 0.4%#24 | – | – | ARC Prize |
Coding
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| SciCode | 47.3%#74 | 51.6%#49 | 53.5%#41 | 55.9%#20 | 56.1%#17 | – | SciCode | |
| WeirdML | 67.2%#35 | – | – | 83.9%#13 | 84.9%#12 | – | WeirdML | |
| FrontierCode | – | – | – | – | – | 43.0%#10 | Cognition | |
| GSO-Bench | – | – | – | – | 40.2%#5 | – | GSO-Bench | |
| LMArena Coding | – | – | – | 1520#28 | – | 1509#48 | LMArena | |
| LMArena WebDev | – | – | – | 1487#43 | 1510#37 | 1458#52 | LMArena | |
| LiveBench Codingnot in index | – | – | – | – | 82.2%#6 | – | LiveBench | |
| SciCode (AA)not in index | – | – | 54.5%#37 | 56.1%#25 | – | 55.8%#27 | Artificial Analysis | |
| LiveCodeBench | – | – | – | – | 85.3%#33 | – | Vals AI | |
| SWE-bench (Vals)not in index | – | – | – | – | 82.6%#19 | – | Vals AI |
Agents & tools
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| Terminal-Bench | – | – | – | – | – | 84.7%#1 | Terminal-Bench | |
| OSWorld-Verified 2.0 | – | – | – | – | 13.0%#9 | – | OSWorld | |
| Remote Labor Index | – | – | – | – | – | 6.3%#3 | Scale AI / CAIS | |
| APEX-Agents | – | – | – | – | 38.5%#12 | 38.5%#12 | Mercor | |
| LMArena Agent | – | – | – | – | 5.3#10 | 3#17 | LMArena | |
| LiveBench Agentic Codingnot in index | – | – | – | – | 54.0%#25 | – | LiveBench | |
| Terminal-Bench Hard | 49.2%#24 | 52.3%#19 | 57.6%#11 | 59.9%#9 | – | 60.6%#7 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 69.3%#153 | 83.9%#102 | 91.8%#56 | 93.0%#47 | – | 93.9%#39 | Artificial Analysis | |
| Terminal-Bench 2.1 (Vals) | – | – | – | 76.4%#13 | – | – | Vals AI | |
| MCP Atlas | – | – | – | – | 75.3%#16 | – | Scale AI SEAL | |
| HiL-Bench | – | – | – | – | – | 39.7%#7 | Scale AI SEAL |
Maths
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | – | – | – | – | 85.3%#8 | – | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | – | – | – | – | 72.5%#14 | – | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | 57.8%#150 | 84.4%#83 | – | – | – | – | Epoch AI Benchmarking Hub | |
| ProofBench | – | – | – | – | 50.0%#17 | – | Vals AI | |
| LiveBench Mathematicsnot in index | – | – | – | – | 95.9%#6 | – | LiveBench | |
| AIME 2026 | – | – | – | – | 100.0%#1 | – | MathArena | |
| HMMT February 2026 | – | – | – | – | 98.5%#1 | – | MathArena | |
| MathArena Apex | – | – | – | – | 80.2%#2 | – | MathArena |
Knowledge
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | – | 63.0%#11 | – | Epoch AI Benchmarking Hub | |
| LiveBench Data Analysisnot in index | – | – | – | – | 81.6%#2 | – | LiveBench | |
| AA-Omniscience | -4.8#116 | 15.1#52 | 18.1#47 | 18.8#46 | – | 20.5#41 | Artificial Analysis | |
| MMLU-Pro | – | – | – | – | 88.1%#21 | – | Vals AI | |
| LegalBench | – | – | – | – | 86.5%#11 | – | Vals AI | |
| CorpFin | – | – | – | – | 68.4%#8 | – | Vals AI | |
| TaxEval | – | – | – | – | 75.0%#24 | – | Vals AI |
Instruction following
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| LiveBench Languagenot in index | – | – | – | – | 87.4%#7 | – | LiveBench | |
| IFBench | 46.1%#193 | 64.3%#108 | 71.0%#63 | 71.6%#52 | – | 75.8%#26 | Artificial Analysis |
Human preference
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| LMArena Text | – | – | – | 1482#19 | – | 1477#24 | LMArena | |
| EQ-Bench 4 | – | – | – | – | – | 1315#4 | EQ-Bench |
Multimodal
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| LMArena Vision | – | – | – | 1294#18 | – | 1296#15 | LMArena | |
| MMMU-Pro | 71.4%#107 | 79.0%#45 | 81.2%#27 | 81.1%#28 | – | 79.9%#38 | Artificial Analysis |
Long context
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| AA-LCR | 64.0%#199 | 81.0%#37 | 83.0%#12 | 84.3%#4 | – | 84.3%#4 | Artificial Analysis |
Composite
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | – | – | – | – | 159.1#8 | Epoch AI Benchmarking Hub | |
| LiveBench | – | – | – | – | 80.2%#6 | – | LiveBench | |
| AA Intelligence Index | 23.2#135 | 30.7#75 | 34.1#60 | 37.3#48 | – | 38.6#44 | Artificial Analysis | |
| Vals Indexnot in index | – | – | – | – | 57.4#16 | – | Vals AI |
Where it wins
Benchmarks where this configuration ranks in the top five of every configuration measured.
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 |
|---|---|---|---|---|---|---|
| Amazon Bedrock (US) | 88 tok/s | 1.62 s | $5.50 | $33.00 | 1.1M | – |
| OpenAI Fast | 80 tok/s | 2.03 s | $12.50 | $75.00 | 1.1M | – |
| OpenAI | 50 tok/s | 2.56 s | $5.00 | $30.00 | 1.1M | – |
| Azure | 36 tok/s | 6.61 s | $5.00 | $30.00 | 1.1M | – |
| Azure (EU) | 31 tok/s | 4.44 s | $5.50 | $33.00 | 1.1M | – |
| Azure (US) | 28 tok/s | 16 s | $5.50 | $33.00 | 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 $0.500 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.011 | $0.0097 | 1.1 min |
| Summarise a 30-page report | 12,000 / 600 | $0.078 | $0.037 | 1.2 min |
| Code edit | 6,000 / 1,500 | $0.075 | $0.055 | 1.3 min |
| Agentic coding session | 60,000 / 4,000 | $0.420 | $0.217 | 1.8 min |
| Structured extraction | 2,000 / 200 | $0.016 | $0.0093 | 1.1 min |
See also
Data as of 9 Sept 2026. Compare these configurations.