OpenAIGPT-5.4 nano
Best configuration ranks #212 of 610 on the BenchLeader Index at 53.9 ±8.5. Last measured 17 Mar 2026. Released 17 Mar 2026.
- Blended price
- $0.463/M
- $0.200 in · $1.25 out
- Output speed
- 169 tok/s
- First answer
- 118 s
- first token 1.67 s
- Context
- 400k
- Overall index54
- Reasoning38
- Agents & tools56
- Knowledge50
- Instruction following73
- Multimodal50
- Long context65
- Composite56
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 | 42.9 | #450 | 153 tok/s | 0.86 s | $0.0005 | Agents & tools 55 · Coding 42 · Composite 44 · Instruction following 36 · Knowledge 46 · Long context 40 · Maths 34 · Multimodal 28 · Reasoning 42 |
| low | 45.3 | #393 | 169 tok/s | 118 s | $0.0005 | Maths 44 · Reasoning 41 |
| medium | 52.6 | #243 | 166 tok/s | 5.94 s | $0.0005 | Agents & tools 63 · Composite 54 · Instruction following 63 · Knowledge 56 · Long context 60 · Multimodal 44 · Reasoning 36 |
| high | 48.3 | #329 | 169 tok/s | 118 s | $0.0005 | Agents & tools 31 · Coding 55 · Human preference 59 · Knowledge 42 · Maths 50 · Multimodal 53 · Reasoning 47 |
| xhigh | – | – | 169 tok/s | 118 s | $0.0005 | Coding 53 · Composite 35 · Reasoning 40 |
| defaultbest | 53.9 | #212 | 169 tok/s | 118 s | $0.0005 | Agents & tools 56 · Composite 56 · Instruction following 73 · Knowledge 50 · Long context 65 · Multimodal 50 · Reasoning 38 |
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 |
|---|---|---|---|---|---|---|---|
| GPQA Diamond | 55.6%#189 | 72.2%#149 | – | 78.5%#122 | – | – | Epoch AI Benchmarking Hub |
| LMArena Hard Prompts | – | – | – | 1422#126 | – | – | LMArena |
| LiveBench Reasoningnot in index | – | – | – | – | 81.1%#37 | – | LiveBench |
| GPQA Diamond (AA)not in index | 55.8%#381 | – | 76.1%#232 | – | – | 81.7%#173 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 4.1%#460 | – | 15.9%#186 | – | – | 28.3%#118 | Artificial Analysis |
| GPQA Diamond (Vals)not in index | – | – | – | 77.5%#74 | – | – | Vals AI |
| Kagi LLM Benchmark | – | – | – | – | – | 39.7%#106 | Kagi LLM Benchmark |
| ARC-AGI-1 | – | 18.3%#158 | 33.0%#143 | 38.2%#135 | 51.5%#122 | – | ARC Prize |
| ARC-AGI-2 | – | 1.5%#151 | 1.9%#144 | 3.6%#133 | 5.7%#117 | – | ARC Prize |
Coding
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source |
|---|---|---|---|---|---|---|---|
| SciCode | – | – | – | – | 46.9%#78 | – | SciCode |
| WeirdML | 38.0%#113 | – | – | 49.2%#69 | – | – | WeirdML |
| LMArena Coding | – | – | – | 1460#114 | – | – | LMArena |
| LiveBench Codingnot in index | – | – | – | – | 70.8%#46 | – | LiveBench |
| SciCode (AA)not in index | – | – | – | – | – | 47.2%#82 | Artificial Analysis |
| LiveCodeBench | – | – | – | 84.0%#40 | – | – | Vals AI |
| IOI | – | – | – | 15.3%#30 | – | – | Vals AI |
| SWE-bench (Vals)not in index | – | – | – | 69.8%#61 | – | – | Vals AI |
Agents & tools
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source |
|---|---|---|---|---|---|---|---|
| APEX-Agents | – | – | – | – | – | 16.9%#45 | Mercor |
| LiveBench Agentic Codingnot in index | – | – | – | – | 46.8%#37 | – | LiveBench |
| Terminal-Bench Hard | 24.2%#137 | – | 33.3%#90 | – | – | 42.4%#44 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 34.8%#230 | – | 52.6%#187 | – | – | 76.0%#130 | Artificial Analysis |
| Terminal-Bench 2.1 (Vals) | – | – | – | 41.6%#56 | – | – | Vals AI |
Maths
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source |
|---|---|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | 4.6%#97 | 20.4%#72 | – | 44.9%#42 | – | – | Epoch AI Benchmarking Hub |
| FrontierMath Tier 4 | – | – | – | 12.2%#47 | – | – | Epoch AI Benchmarking Hub |
| OTIS Mock AIME | 46.7%#169 | 68.9%#126 | – | 87.8%#68 | – | – | Epoch AI Benchmarking Hub |
| ProofBench | – | – | – | 5.0%#55 | – | – | Vals AI |
| LiveBench Mathematicsnot in index | – | – | – | – | 91.0%#21 | – | LiveBench |
| AIME (Vals) | – | – | – | 88.8%#28 | – | – | Vals AI |
Knowledge
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source |
|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | 11.7%#71 | – | – | Epoch AI Benchmarking Hub |
| LiveBench Data Analysisnot in index | – | – | – | – | 67.6%#43 | – | LiveBench |
| AA-Omniscience | -37.5#245 | – | -18.0#177 | – | – | -29.5#212 | Artificial Analysis |
| MMLU-Pro | – | – | – | 77.2%#102 | – | – | Vals AI |
| LegalBench | – | – | – | 77.9%#96 | – | – | Vals AI |
| CorpFin | – | – | – | 61.2%#56 | – | – | Vals AI |
| TaxEval | – | – | – | 67.4%#104 | – | – | Vals AI |
Instruction following
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source |
|---|---|---|---|---|---|---|---|
| LiveBench Languagenot in index | – | – | – | – | 62.5%#52 | – | LiveBench |
| IFBench | 32.7%#323 | – | 64.4%#106 | – | – | 75.9%#24 | Artificial Analysis |
Human preference
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source |
|---|---|---|---|---|---|---|---|
| LMArena Text | – | – | – | 1402#129 | – | – | LMArena |
Multimodal
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source |
|---|---|---|---|---|---|---|---|
| LMArena Vision | – | – | – | 1198#74 | – | – | LMArena |
| MMMU-Pro | 43.8%#223 | – | 59.5%#181 | – | – | 65.4%#142 | Artificial Analysis |
Long context
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source |
|---|---|---|---|---|---|---|---|
| AA-LCR | 29.7%#324 | – | 67.3%#182 | – | – | 76.7%#96 | Artificial Analysis |
Composite
| Benchmark | no reasoning | low | medium | high | xhigh | default | Source |
|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | – | – | – | – | 145.9#64 | Epoch AI Benchmarking Hub |
| LiveBench | – | – | – | – | 69.6%#42 | – | LiveBench |
| AA Intelligence Index | 11.7#281 | – | 20.0#174 | – | – | 21.2#162 | Artificial Analysis |
| Vals Indexnot in index | – | – | – | 33#39 | – | – | 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 |
|---|---|---|---|---|---|---|
| OpenAI | 69 tok/s | 1.21 s | $0.200 | $1.25 | 400k | – |
| OpenAI Flex | 58 tok/s | 1.60 s | $0.100 | $0.625 | 400k | – |
| Azure | 34 tok/s | 1.74 s | $0.200 | $1.25 | 400k | – |
| Azure (US) | 24 tok/s | 2.60 s | $0.220 | $1.38 | 400k | – |
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 | $0.0005 | $0.0004 | 2.0 min |
| Summarise a 30-page report | 12,000 / 600 | $0.0032 | $0.0015 | 2.0 min |
| Code edit | 6,000 / 1,500 | $0.0031 | $0.0023 | 2.1 min |
| Agentic coding session | 60,000 / 4,000 | $0.017 | $0.0089 | 2.4 min |
| Structured extraction | 2,000 / 200 | $0.0007 | $0.0004 | 2.0 min |
See also
Data as of 9 Sept 2026. Compare these configurations.