OpenAIGPT-4.1
GPT-4.1 is an OpenAI proprietary model, released 14 Apr 2025. Its best configuration ranks #428 of 785 on the BenchLeader Index at 48.3 ±4.0, in the lower half. The ± is the point: 303 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 human preference (59) and lowest in knowledge (39). At $3.50 per million tokens blended it is among the most expensive ranked models. Output speed of 161 tokens per second puts it in the fastest quarter, with a first answer in 0.9 s. It has been measured at 2 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 22 Oct 2026.
- Blended price
- $3.50/M
- $2.00 in · $8.00 out
- Output speed
- 161 tok/s
- measured by Artificial Analysis
- First answer
- 0.89 s
- first token 1.01 s
- Context
- 1.0M
- Full answer
- 4.00 s
- median, reasoning included
- Cost per run
- –
- Released
- 14 Apr 2025
- Overall index48
- Reasoning39
- Coding43
- Agents & tools58
- Maths48
- Knowledge39
- Instruction following45
- Human preference59
- Multimodal51
- Long context57
- Composite44
Versions
OpenAI has shipped 13 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-5.6 | 9 Jul 2026 | – | – |
| GPT-5.5xhigh | 23 Apr 2026 | 66.1 | #35 |
| GPT-5.4xhigh | 5 Mar 2026 | 64.7 | #45 |
| GPT-5.2xhigh | 11 Dec 2025 | 60.7 | #104 |
| GPT-5.1high | 12 Nov 2025 | 57.7 | #170 |
| GPT-5high | 7 Aug 2025 | 57.9 | #166 |
| GPT-4.1this page | 14 Apr 2025 | 48.3 | #428 |
| GPT-4.5 | 27 Feb 2025 | 49.7 | #388 |
| GPT 4 0125 | 25 Jan 2024 | – | – |
| GPT-4 1106 | 6 Nov 2023 | – | – |
| GPT 4 0613 | 13 Jun 2023 | 39.7 | #677 |
| GPT-4 | 14 Mar 2023 | 42.4 | #596 |
| GPT-3.5 | – | – | – |
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 |
|---|---|---|---|---|---|---|
| high | 46.7 | #477 | 161 tok/s | 0.89 s | $0.0032 | Coding 38 · Knowledge 55 · Maths 43 |
| not statedbest | 48.3 | #428 | 161 tok/s | 0.89 s | $0.0032 | Agents & tools 58 · Coding 43 · Composite 44 · Human preference 59 · Instruction following 45 · Knowledge 39 · Long context 57 · Maths 48 · Multimodal 51 · Reasoning 39 |
How its index has moved
8 Sept 2026 to 21 Oct 2026- not stated
- high
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 | high | not stated | Source | Trend |
|---|---|---|---|---|
| GPQA Diamondeffort not stated | – | 66.9%#177 | Epoch AI Benchmarking Hub | |
| Humanity's Last Exameffort not stated | – | 5.4%#42 | Scale AI / CAIS | |
| SimpleBencheffort not stated | – | 27.0%#72 | SimpleBench | |
| Arena Hard Promptseffort not stated | – | 1432#134 | Arena (LMArena) | |
| GPQA Diamond (AA)not in indexeffort not stated | – | 66.6%#330 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in indexeffort not stated | – | 4.2%#530 | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | 65.4%#97 | – | Vals AI | |
| Kagi LLM Benchmarkeffort not stated | – | 52.3%#73 | Kagi LLM Benchmark | |
| GPQA Diamond (HELM)not in indexeffort not stated | – | 65.9%#18 | HELM Capabilities | |
| ARC-AGI-1effort not stated | – | 5.5%#243 | ARC Prize | |
| ARC-AGI-2effort not stated | – | 0.4%#239 | ARC Prize | |
| CritPteffort not stated | – | 0.0%#315 | Artificial Analysis | |
| Chess Puzzlesnot in indexeffort not stated | – | 3.0%#135 | Epoch AI Benchmarking Hub | |
| LMCAnot in indexeffort not stated | – | 25.6%#148 | Epoch AI Benchmarking Hub | |
| DTBenchnot in indexeffort not stated | – | 68.3%#161 | Epoch AI Benchmarking Hub | |
| ForecastBenchnot in indexeffort not stated | – | 62.3%#57 | ForecastBench |
Coding
| Benchmark | high | not stated | Source | Trend |
|---|---|---|---|---|
| SWE-bench Verified (Epoch)effort not stated | – | 48.5%#32 | Epoch AI Benchmarking Hub | |
| WeirdMLeffort not stated | – | 39.0%#109 | WeirdML | |
| Arena Codingeffort not stated | – | 1456#140 | Arena (LMArena) | |
| LiveCodeBench | 54.7%#108 | – | Vals AI | |
| Aider Polygloteffort not stated | – | 52.4%#24 | Aider polyglot leaderboard | |
| SWE-bench Verified (bash only)effort not stated | – | 39.6%#36 | SWE-bench | |
| SWE-bench Verified (any scaffold)not in indexeffort not stated | – | 39.6%#47 | SWE-bench | |
| CadEvalnot in indexeffort not stated | – | 42.0%#8 | Epoch AI Benchmarking Hub | |
| ALE-Benchnot in indexeffort not stated | – | 558.1#102 | Epoch AI Benchmarking Hub |
Agents & tools
| Benchmark | high | not stated | Source | Trend |
|---|---|---|---|---|
| Terminal-Bench Hardeffort not stated | – | 13.6%#201 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in indexeffort not stated | – | 47.1%#203 | Artificial Analysis | |
| BFCL Overalleffort not stated | – | 54.0%#18 | Berkeley Function Calling Leaderboard | |
| Terminal-Bench 2.0 (Vals)not in index | 14.6%#60 | – | Vals AI | |
| BrowseComp-Plusnot in indexeffort not stated | – | 15.3%#13 | BrowseComp-Plus | |
| BrowseCompnot in indexeffort not stated | – | 3.5%#6 | OpenAI |
Maths
| Benchmark | high | not stated | Source | Trend |
|---|---|---|---|---|
| FrontierMath Tiers 1–3effort not stated | – | 6.0%#109 | Epoch AI Benchmarking Hub | |
| OTIS Mock AIMEeffort not stated | – | 38.3%#199 | Epoch AI Benchmarking Hub | |
| MATH Level 5effort not stated | – | 83.0%#31 | Epoch AI Benchmarking Hub | |
| AIME (Vals) | 39.6%#67 | – | Vals AI | |
| MGSM | 87.7%#55 | – | Vals AI | |
| Omni-MATH (HELM)effort not stated | – | 47.1%#21 | HELM Capabilities | |
| MATH 500not in index | 87.2%#31 | – | Vals AI | |
| Arena Mathseffort not stated | – | 1375#177 | Arena (LMArena) |
Knowledge
| Benchmark | high | not stated | Source | Trend |
|---|---|---|---|---|
| SimpleQA Verifiedeffort not stated | – | 31.1%#65 | Epoch AI Benchmarking Hub | |
| AA-Omniscienceeffort not stated | – | -39.6#321 | Artificial Analysis | |
| MMLU-Pro | 80.5%#83 | – | Vals AI | |
| LegalBench | 83.1%#58 | – | Vals AI | |
| CorpFin | 63.0%#46 | – | Vals AI | |
| TaxEval | 75.1%#23 | – | Vals AI | |
| MedQA | 91.2%#39 | – | Vals AI | |
| PRBench Financeeffort not stated | – | 34.3%#34 | Scale AI SEAL | |
| PRBench Legaleffort not stated | – | 36.5%#38 | Scale AI SEAL | |
| MultiNRCeffort not stated | – | 21.2%#33 | Scale AI SEAL | |
| MMLU-Pro (HELM)not in indexeffort not stated | – | 81.1%#16 | HELM Capabilities | |
| CaseLaw v2not in index | 69.9%#3 | – | Vals AI | |
| AA-Omniscience: accuracynot in indexeffort not stated | – | 27.8%#202 | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in indexeffort not stated | – | 6.7%#499 | Artificial Analysis |
Instruction following
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| IFBench | 43.0%#230 | Artificial Analysis | |
| MultiChallenge | 39.4%#31 | Scale AI SEAL | |
| IFEval (HELM) | 83.8%#28 | HELM Capabilities | |
| WildBench (HELM) | 85.4%#9 | HELM Capabilities | |
| Arena Instruction Followingnot in index | 1403#132 | Arena (LMArena) |
Human preference
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Text | 1415#132 | Arena (LMArena) | |
| Arena Creative Writingnot in index | 1403#103 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1430#114 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1426#126 | Arena (LMArena) |
Multimodal
| Benchmark | high | not stated | Source | Trend |
|---|---|---|---|---|
| Arena Visioneffort not stated | – | 1210#85 | Arena (LMArena) | |
| MMMU-Proeffort not stated | – | 61.2%#207 | Artificial Analysis | |
| VISTAeffort not stated | – | 45.3%#26 | Scale AI SEAL | |
| MMMU-Pro (Vals)not in index | 72.4%#62 | – | Vals AI | |
| MortgageTaxnot in index | 65.9%#37 | – | Vals AI | |
| VTBnot in indexeffort not stated | – | 5.5%#13 | Scale AI SEAL | |
| GeoBenchnot in indexeffort not stated | – | 72.0%#11 | GeoBench |
Long context
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Fiction.LiveBench 120k | 62.5%#13 | Fiction.live | |
| AA-LCR | 68.3%#235 | Artificial Analysis |
Composite
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Epoch Capabilities Indexnot in index | 136.7#126 | Epoch AI Benchmarking Hub | |
| AA Intelligence Index v4.3.2 | 12.7#326 | Artificial Analysis | |
| HELM Capabilities meannot in index | 72.7%#17 | HELM Capabilities |
Safety & honesty
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| FORTRESSnot in index | 53.0%#11 | Scale AI SEAL | |
| MASKnot in indexpossible contamination | 51.1%#48 | Scale AI SEAL |
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 | 194 tok/s | 0.90 s | $2.00 | $8.00 | 1.0M | – |
| OpenAI | 80 tok/s | 0.78 s | $2.00 | $8.00 | 1.0M | – |
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 | $0.0032 | $0.0028 | 2.8 s |
| Summarize a 30-page report | 12,000 / 600 | $0.029 | $0.015 | 4.6 s |
| Code edit | 6,000 / 1,500 | $0.024 | $0.017 | 10.2 s |
| Agentic coding session | 60,000 / 4,000 | $0.152 | $0.085 | 25.8 s |
| Structured extraction | 2,000 / 200 | $0.0056 | $0.0034 | 2.1 s |
More from OpenAI
See also
Questions people ask
- What are GPT-4.1's benchmark scores?
- As of 25 Oct 2026, GPT-4.1 scores 48.3 ±4.0 on the BenchLeader Index, strongest in human preference (59) and weakest in knowledge (39). That rests on 36 independent benchmark results, and every score on this page links to its publisher.
- How does GPT-4.1 rank?
- GPT-4.1 is #428 of 785 ranked configurations on 25 Oct 2026. 303 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-4.1 cost per 1M tokens?
- As of 25 Oct 2026, GPT-4.1 costs $2.00 per 1M input tokens and $8.00 per 1M output tokens, $3.50 blended (three input tokens per output token).
- When was GPT-4.1 released?
- OpenAI released GPT-4.1 on 14 Apr 2025. BenchLeader has results for it at 2 reasoning-effort settings.
Data as of 25 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “GPT-4.1: benchmarks, pricing, speed and rank”, data as of 25 Oct 2026, https://www.benchleader.com/models/gpt-4-1.