OpenAIGPT-5.4
Best configuration ranks #46 of 610 on the BenchLeader Index at 63.4 ±6.2. Last measured 8 Sept 2026. Released 5 Mar 2026.
- Blended price
- $5.63/M
- $2.50 in · $15.00 out
- Output speed
- 130 tok/s
- First answer
- 126 s
- first token 3.21 s
- Context
- 1.1M
- Overall index63
- Reasoning62
- Coding52
- Agents & tools66
- Knowledge67
- Instruction following72
- Human preference65
- Multimodal64
- Long context68
- 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 | 53.7 | #216 | 101 tok/s | 0.83 s | $0.0055 | Agents & tools 67 · Coding 55 · Composite 52 · Instruction following 50 · Knowledge 57 · Long context 55 · Maths 49 · Multimodal 55 · Reasoning 55 |
| low | 60.4 | #88 | 106 tok/s | 1.68 s | $0.0055 | Agents & tools 71 · Composite 64 · Instruction following 65 · Knowledge 66 · Long context 65 · Maths 61 · Multimodal 63 · Reasoning 54 |
| medium | 56.0 | #168 | 130 tok/s | 126 s | $0.0061 | Coding 51 · Maths 66 · Reasoning 62 |
| high | 58.8 | #115 | 130 tok/s | 126 s | $0.0061 | Agents & tools 51 · Coding 58 · Human preference 68 · Knowledge 53 · Maths 67 · Multimodal 66 · Reasoning 62 |
| xhigh | 61.3 | #67 | 130 tok/s | 126 s | $0.0061 | Agents & tools 54 · Coding 68 · Composite 60 · Knowledge 59 · Maths 67 · Reasoning 68 |
| thinking | – | – | 130 tok/s | 126 s | $0.0061 | Knowledge 70 · Multimodal 65 |
| defaultbest | 63.4 | #46 | 130 tok/s | 126 s | $0.0055 | Agents & tools 66 · Coding 52 · Composite 78 · Human preference 65 · Instruction following 72 · Knowledge 67 · Long context 68 · Multimodal 64 · Reasoning 62 |
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 | thinking | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| GPQA Diamond | 74.8%#139 | 84.8%#81 | 88.9%#45 | 89.9%#39 | 93.3%#15 | – | – | Epoch AI Benchmarking Hub | |
| Humanity's Last Exam | – | – | – | – | 36.2%#6 | – | – | Scale AI / CAIS | |
| LMArena Hard Prompts | – | – | – | 1498#27 | – | – | 1488#44 | LMArena | |
| LiveBench Reasoningnot in index | – | – | – | – | 88.1%#17 | – | – | LiveBench | |
| GPQA Diamond (AA)not in index | 74.8%#245 | 87.1%#102 | – | – | – | – | 92.0%#41 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 11.3%#232 | 30.8%#97 | – | – | – | – | 43.7%#31 | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | – | – | – | – | 91.7%#18 | – | – | Vals AI | |
| Kagi LLM Benchmark | – | – | – | – | – | – | 63.8%#40 | Kagi LLM Benchmark | |
| ARC-AGI-1 | – | 68.2%#98 | 86.2%#68 | 92.7%#34 | 93.7%#31 | – | – | ARC Prize | |
| ARC-AGI-2 | – | 29.2%#90 | 55.4%#65 | 67.5%#41 | 74.0%#32 | – | – | ARC Prize | |
| ARC-AGI-3 | – | – | – | 0.2%#30 | – | – | – | ARC Prize |
Coding
| Benchmark | no reasoning | low | medium | high | xhigh | thinking | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| SWE-bench Verified (Epoch) | – | – | – | 76.9%#8 | – | – | – | Epoch AI Benchmarking Hub | |
| SciCode | – | – | – | – | 56.6%#12 | – | – | SciCode | |
| WeirdML | 57.4%#50 | – | – | – | 77.7%#21 | – | – | WeirdML | |
| GSO-Bench | – | – | – | 25.5%#11 | 31.4%#8 | – | – | GSO-Bench | |
| LMArena Coding | – | – | – | 1520#27 | – | – | 1514#41 | LMArena | |
| LMArena WebDev | – | – | 1443#56 | 1463#50 | – | – | 1386#77 | LMArena | |
| LiveBench Codingnot in index | – | – | – | – | 77.5%#26 | – | – | LiveBench | |
| LiveCodeBench | – | – | – | – | 84.1%#39 | – | – | Vals AI | |
| IOI | – | – | – | – | 67.8%#6 | – | – | Vals AI | |
| SWE-bench (Vals)not in index | – | – | – | – | 78.2%#30 | – | – | Vals AI | |
| SWE-Bench Pro | – | – | – | – | 59.1%#2 | – | – | Scale AI SEAL |
Agents & tools
| Benchmark | no reasoning | low | medium | high | xhigh | thinking | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Terminal-Bench | – | – | – | – | – | – | 81.8%#2 | Terminal-Bench | |
| APEX-Agents | – | – | – | – | 36.0%#16 | – | 34.9%#18 | Mercor | |
| LMArena Agent | – | – | – | 1.6#21 | – | – | – | LMArena | |
| LiveBench Agentic Codingnot in index | – | – | – | – | 53.8%#26 | – | – | LiveBench | |
| Terminal-Bench Hard | 37.9%#61 | 43.2%#39 | – | – | – | – | 57.6%#11 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 36.0%#228 | 74.6%#135 | – | – | – | – | 87.1%#79 | Artificial Analysis | |
| MCP Atlas | – | – | – | – | 70.6%#17 | – | – | Scale AI SEAL | |
| HiL-Bench | – | – | – | – | – | – | 9.7%#15 | Scale AI SEAL |
Maths
| Benchmark | no reasoning | low | medium | high | xhigh | thinking | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | – | – | – | – | 78.6%#12 | – | – | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | – | – | – | – | 49.0%#19 | – | – | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | 57.8%#150 | 84.4%#83 | 95.6%#29 | 97.8%#17 | 95.3%#37 | – | – | Epoch AI Benchmarking Hub | |
| ProofBench | – | – | – | – | 56.0%#13 | – | – | Vals AI | |
| LiveBench Mathematicsnot in index | – | – | – | – | 94.2%#11 | – | – | LiveBench | |
| AIME (Vals) | – | – | – | – | 96.7%#5 | – | – | Vals AI | |
| AIME 2026 | – | – | – | – | 99.2%#3 | – | – | MathArena | |
| HMMT February 2026 | – | – | – | – | 97.7%#2 | – | – | MathArena | |
| MathArena Apex | – | – | – | – | 54.2%#6 | – | – | MathArena |
Knowledge
| Benchmark | no reasoning | low | medium | high | xhigh | thinking | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | – | 45.1%#32 | – | – | Epoch AI Benchmarking Hub | |
| LiveBench Data Analysisnot in index | – | – | – | – | 79.3%#9 | – | – | LiveBench | |
| AA-Omniscience | -15.7#165 | 4.8#73 | – | – | – | – | 5.8#69 | Artificial Analysis | |
| MMLU-Pro | – | – | – | – | 87.5%#27 | – | – | Vals AI | |
| LegalBench | – | – | – | – | 86.0%#13 | – | – | Vals AI | |
| CorpFin | – | – | – | – | 65.3%#32 | – | – | Vals AI | |
| TaxEval | – | – | – | – | 74.0%#42 | – | – | Vals AI | |
| MedQA | – | – | – | – | 96.1%#5 | – | – | Vals AI | |
| PRBench Finance | – | – | – | 45.6%#17 | – | – | – | Scale AI SEAL | |
| PRBench Legal | – | – | – | 44.4%#15 | – | – | – | Scale AI SEAL | |
| MultiNRC | – | – | – | – | – | 58.3%#7 | – | Scale AI SEAL |
Instruction following
| Benchmark | no reasoning | low | medium | high | xhigh | thinking | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| LiveBench Languagenot in index | – | – | – | – | 82.6%#18 | – | – | LiveBench | |
| IFBench | 48.4%#180 | 65.9%#100 | – | – | – | – | 74.0%#37 | Artificial Analysis |
Human preference
| Benchmark | no reasoning | low | medium | high | xhigh | thinking | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| LMArena Text | – | – | – | 1477#25 | – | – | 1466#46 | LMArena | |
| EQ-Bench 4 | – | – | – | – | – | – | 1272#7 | EQ-Bench |
Multimodal
| Benchmark | no reasoning | low | medium | high | xhigh | thinking | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| LMArena Vision | – | – | – | 1298#14 | – | – | 1293#20 | LMArena | |
| MMMU-Pro | 70.6%#111 | 78.0%#56 | – | – | – | – | 78.4%#54 | Artificial Analysis | |
| VISTA | – | – | – | – | – | 50.9%#8 | – | Scale AI SEAL |
Long context
| Benchmark | no reasoning | low | medium | high | xhigh | thinking | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| AA-LCR | 58.3%#225 | 76.7%#96 | – | – | – | – | 82.0%#25 | Artificial Analysis |
Composite
| Benchmark | no reasoning | low | medium | high | xhigh | thinking | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | – | – | – | – | – | 156.9#13 | Epoch AI Benchmarking Hub | |
| LiveBench | – | – | – | – | 78.0%#12 | – | – | LiveBench | |
| AA Intelligence Index | 18.2#192 | 27.6#97 | – | – | – | – | 39.0#42 | Artificial Analysis |
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 |
|---|---|---|---|---|---|---|
| OpenAI | 53 tok/s | 2.13 s | $2.50 | $15.00 | 1.1M | – |
| Azure | 51 tok/s | 4.28 s | $2.50 | $15.00 | 1.1M | – |
| Amazon Bedrock (US) | 32 tok/s | 1.33 s | $2.75 | $16.50 | 1.1M | – |
| OpenAI Flex | 2 tok/s | 12 s | $1.25 | $7.50 | 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 $0.250 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0055 | $0.0048 | 2.1 min |
| Summarise a 30-page report | 12,000 / 600 | $0.039 | $0.019 | 2.2 min |
| Code edit | 6,000 / 1,500 | $0.037 | $0.027 | 2.3 min |
| Agentic coding session | 60,000 / 4,000 | $0.210 | $0.109 | 2.6 min |
| Structured extraction | 2,000 / 200 | $0.0080 | $0.0046 | 2.1 min |
See also
Data as of 9 Sept 2026. Compare these configurations.