Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
OpenAI · Proprietary · rank 46 · bench-align-v5
GPT-5.4 is built for agent work. It ranks in the top quarter for agent work (top 15% of 127 scored models) and instruction following (top 17% of 159 scored models), by BenchLM's published scores. It accepts text, images and PDFs and produces text.
GPT-5.4 by OpenAI scores 62.7 overall, costs $2.50 per 1M input tokens and $15.00 per 1M output tokens, and carries a context window of 1.05M tokens — observed 2026-11-21.
Every published price component, including cache reads and writes.
Lifecycle events the source associates with this model, plus what this profile does not claim.
That is not evidence the model has no lifecycle plan — only that this source published none.
Seven axes from the ranking source. A missing axis is drawn as a gap.
Each axis shows the ranking source’s score and, beside it, BenchLM’s separate verified score for that axis. The two are published side by side and are not interchangeable.
Measured values with the date they were observed. Nothing is inferred from a sibling model or a provider claim.
Time to first token: latency to the first answer chunk (Artificial Analysis, via BenchLM). Reasoning models include thinking time, so values can run to tens or hundreds of seconds.
Evidence key: Observed
| Measurement | Value | Observed | Last good | Evidence |
|---|---|---|---|---|
| Time to first token | 157.51 s | 2026-11-21 | 2026-11-21 | |
| Throughput | 89 tok/s | 2026-11-21 | 2026-11-21 |
Regional or per-endpoint measurements appear only when the API supplies them; none are modelled here.
| Batch output / 1M tokens (OpenRouter batch) (base rate) | $7.50 |
| Batch cache read / 1M tokens (OpenRouter batch) (base rate) | $0.13 |
Batch API rate (asynchronous jobs), from OpenRouter's batch listing.
The standard input, output and cache rates above are the base rate. models.dev publishes higher rates for requests whose prompt exceeds the thresholds below. The batch rates have their own long-context tiers, from OpenRouter's batch listing, shown separately. The cost simulator prices a standard-rate request whose whole prompt is above a threshold entirely at the highest of these tiers it exceeds, and a batch request only at a batch tier; it never applies these tiers to batch rates. The cost example on this page uses the base rate: its 1,200-token prompt is below every threshold.
The batch rates above are the base batch rate. OpenRouter's batch listing, the same listing as those rates, publishes higher batch rates for requests whose total prompt is above the thresholds below. The cost simulator prices a batch request whose whole prompt is above a threshold entirely at the highest batch tier it exceeds.
10 conversations per day × 8 messages × 22 active days, 1200 input and 400 output tokens per message, no cache. This uses the same calculator as the cost simulator, so an unavailable applicable rate makes the example unavailable too.
Plans from the model's maker, and the API providers that list this exact model today. Links leave ModelChoice; each provider sets its own prices and limits.
| Benchmark | Value | Rank | Tasks | Difficulty | Provenance |
|---|---|---|---|---|---|
| GPQA Graduate-Level Google-Proof Q&A | 92.8 | #16 | 448 questions | Graduate level | GPQA: A Graduate-Level Google-Proof Q&A Benchmark |
| GPQA-D GPQA Diamond | 92.8 | #17 | Graduate-level science questions | Graduate level | GPQA: A Graduate-Level Google-Proof Q&A Benchmark |
| HLE Humanity's Last Exam | 52.1 | #23 | Expert-level questions | Frontier expert level | Humanity's Last Exam |
| Artificial Analysis Intelligence Index Artificial Analysis Intelligence Index | 39.0 | #40 | Cross-benchmark intelligence index | Display-only external reference | Artificial Analysis |
| AA-GPQA Diamond Artificial Analysis GPQA Diamond | 92.0 | #23 | Graduate-level science questions | Graduate-level science reasoning | Artificial Analysis GPQA Diamond Benchmark Leaderboard |
| AA-HLE Artificial Analysis Humanity's Last Exam | 43.7 | #31 | Expert-level questions | Frontier expert reasoning | Artificial Analysis Humanity's Last Exam Benchmark Leaderboard |
| AA-Omniscience Index Artificial Analysis Omniscience Index | 5.8 | #56 | Knowledge questions | Broad factual knowledge | AA-Omniscience: Knowledge and Hallucination Benchmark |
| AA-Omniscience Accuracy Artificial Analysis Omniscience Accuracy | 50.8 | #23 | Knowledge questions | Broad knowledge | Artificial Analysis model benchmarks |
| AA-Omniscience Hallucination Rate Artificial Analysis Omniscience Hallucination Rate | 91.7 | #25 | Knowledge questions | Factuality | Artificial Analysis model benchmarks |
| HealthBench Hard HealthBench Hard | 40.1 | #2 | 1,000 health prompts | Advanced health reasoning | Muse Spark Eval Methodology |
| HealthBench Professional HealthBench Professional | 48.1 | #21 | Clinician chat tasks | Professional clinical workflows | HealthBench Professional: Evaluating Large Language Models on Real Clinician Chats |
| MedXpertQA (Text) MedXpertQA Text | 59.6 | #2 | 2,450 medical multiple-choice questions | Professional medical knowledge | Muse Spark Eval Methodology |
| HLE w/o tools Humanity's Last Exam without tools | 39.8 | #32 | Expert-level questions | Frontier expert level | Introducing GPT-5.4 mini and nano |
| GPQA Diamond (Vals) GPQA Diamond, Vals AI run | 91.7 | #19 | Graduate-level science questions | Expert reasoning | Vals AI GPQA Diamond, Vals AI run leaderboard |
| MMLU-Pro (Vals) MMLU-Pro, Vals AI run | 87.5 | #25 | Academic multiple-choice questions | Broad academic knowledge | Vals AI MMLU-Pro, Vals AI run leaderboard |
| Benchmark | Value | Rank | Tasks | Difficulty | Provenance |
|---|---|---|---|---|---|
| Terminal-Bench 2.0 Terminal-Bench 2.0 | 75.1 | #17 | Terminal-based software tasks | Professional software engineering | Terminal-Bench 2.0 |
| LiveCodeBench Pro LiveCodeBench Pro | 87.5 | #3 | Quarter-specific contest programming sets | High-end contest programming | LiveCodeBench Pro: How Do Olympiad Medalists Judge LLMs in Competitive Programming? |
| SWE-bench Pro SWE-bench Pro | 57.7 |
| Benchmark | Value | Rank | Tasks | Difficulty | Provenance |
|---|---|---|---|---|---|
| FrontierMath v2 (Tiers 1-3) FrontierMath v2 Tiers 1-3 | 78.596 | #24 | 285 private advanced mathematics problems | From olympiad-plus to early research mathematics | FrontierMath v2 benchmark hub |
| FrontierMath v2 (Tier 4) FrontierMath v2 Tier 4 | 49.000 | #24 | 41 private extreme-difficulty mathematics problems | Research-level mathematics requiring hours or days of expert work | FrontierMath Tier 4 v2 leaderboard |
| Benchmark | Value | Rank | Tasks | Difficulty | Provenance |
|---|---|---|---|---|---|
| ARC-AGI-1 ARC-AGI-1 Semi-Private Evaluation | 93.67 | #21 | Semi-private ARC-AGI-1 evaluation set | Abstract visual reasoning | ARC Prize leaderboard |
| ARC-AGI-2 Abstraction and Reasoning Corpus for AGI v2 | 73.95 | #24 | Visual pattern completion and abstract reasoning | Expert-level — hardest public reasoning benchmark | ARC-AGI-2: A Harder General Intelligence Benchmark |
| ARC-AGI-3 Abstraction and Reasoning Corpus for AGI v3 | 0.21 |
| Benchmark | Value | Rank | Tasks | Difficulty | Provenance |
|---|---|---|---|---|---|
| AA-IFBench Artificial Analysis IFBench | 73.9 | #27 | Verifiable instruction constraints | Instruction precision | Artificial Analysis IFBench Benchmark Leaderboard |
| Benchmark | Value | Rank | Tasks | Difficulty | Provenance |
|---|---|---|---|---|---|
| AA Tau3 Banking Artificial Analysis Tau3-Banking | 39.6 | #16 | Banking tool-use workflows | Agentic banking workflows | Artificial Analysis Tau3-Banking Benchmark Leaderboard |
| Terminal-Bench 2.0 Terminal-Bench 2.0 | 75.1 | #18 | Terminal-based software tasks | Professional software engineering | Terminal-Bench 2.0 |
| BrowseComp BrowseComp | 82.7 |
| Benchmark | Value | Rank | Tasks | Difficulty | Provenance |
|---|---|---|---|---|---|
| MMMU-Pro Massive Multi-discipline Multimodal Understanding Pro | 81.2 | #10 | Multimodal academic reasoning | Frontier multimodal | MMMU-Pro: A More Robust Multi-discipline Multimodal Understanding Benchmark |
| AA-MMMU-Pro Artificial Analysis MMMU-Pro | 78.4 | #39 | Multimodal academic reasoning | Frontier multimodal | Artificial Analysis MMMU-Pro Benchmark Leaderboard |
| Design Arena Website Design Arena Website Elo | 1207 |
| Benchmark | Value | Rank | Tasks | Difficulty | Provenance |
|---|---|---|---|---|---|
| Chess Puzzles Chess Puzzles (Epoch AI) | 48.0 | #21 | 100 programmatically generated chess positions with a single best move | Novel chess positions, single best move | Chess Puzzles (Epoch AI) |
| Mystery Game Puzzles Mystery Game Puzzles (Epoch AI) | 37.0 | #24 | 100 single-best-move positions from an undisclosed game | Unfamiliar game positions, single best move | Mystery Game Puzzles (Epoch AI) |
| Furniture Assembly Furniture Assembly Benchmark (Epoch AI) | 37.5 |
People's votes between two anonymous answers on arena.ai, one board per task. LMArena's own ratings and ranks, one line per model version it rates; not part of this site's capability scores or rank.
Source: LMArena, CC BY 4.0 (licence), leaderboard data read once a day.
| #42 |
| 1,865 repository problems |
| Long-horizon professional engineering |
| SWE-Bench Pro: Can AI Agents Solve Long-Horizon Software Engineering Tasks? |
| Vibe Code Bench Vibe Code Bench v1.1 | 67.42 | #5 | End-to-end web application builds | End-to-end software delivery | Vibe Code Bench: Evaluating AI Models on End-to-End Web Application Development |
| React Native Evals React Native Evals | 85.3 | #3 | React Native app implementation tasks | Production mobile app engineering | React Native Evals |
| AA Coding Index Artificial Analysis Coding Index | 71.0 | #24 | Cross-benchmark coding index | Display-only external reference | Artificial Analysis model leaderboards |
| LiveCodeBench (Vals) LiveCodeBench, Vals AI run | 84.1 | #39 | Competitive programming problems (easy, medium, hard) | Frontier coding | Vals AI LiveCodeBench, Vals AI run leaderboard |
| SWE-bench (Vals) SWE-bench, Vals AI run | 78.2 | #30 | Real repository issues by human time bucket | Frontier coding agents | Vals AI SWE-bench, Vals AI run leaderboard |
| PostTrainBench v1.1 PostTrainBench v1.1 | 19.0 | #15 | Post-training Qwen3 1.7B and 4B, SmolLM3 3B, and Gemma3 4B | Frontier agent evaluation | PostTrainBench v1.1 leaderboard and methodology |
| PostTrainBench v1 PostTrainBench v1 (archived) | 20.2 | #8 | Post-training four base models on seven benchmarks | Autonomous ML engineering | PostTrainBench v1 leaderboard (archived) |
| PostTrainBench PostTrainBench v1.2 | 17.8 | #23 | Post-training four base models on six benchmarks | Autonomous ML engineering | PostTrainBench |
| #22 |
| Interactive game-like tasks with hidden rules |
| Frontier agentic reasoning |
| ARC-AGI-3: A New Challenge for Frontier Agentic Intelligence |
| AA-LCR Artificial Analysis Long Context Reasoning | 82.0 | #29 | Long-context reasoning tasks | Long-context reasoning | Artificial Analysis model benchmarks |
| CritPt Critical Physics Tasks | 23.4 | #23 | Research-level physics questions | Research-level physics reasoning | CritPt Benchmark Leaderboard |
| BioBasicsBench BioBasicsBench: everyday wet-lab science | 33.2 | #6 | 100 routine wet-lab analysis tasks (88 multimodal) across five domains | Routine wet-lab data interpretation | Measuring how AI agents handle everyday wet-lab science |
| #28 |
| Research questions requiring browsing |
| Hard web research |
| BrowseComp |
| GDPval-AA GDPval-AA | 1307 | #54 | Agentic real-world work tasks | Professional agentic workflows | DeepSeek-V4 Technical Report |
| GDPval-AA GDPval-AA normalized | 37.3 | #49 | Economically valuable tasks | Professional agentic workflows | Artificial Analysis model benchmarks |
| APEX-Agents-AA APEX-Agents-AA | 33.3 | #10 | 452 professional-services agent tasks | Long-horizon workplace agent tasks | APEX-Agents-AA Benchmark Leaderboard |
| Gert Labs Gert Labs Composite Game Benchmark | 64.89 | #4 | Novel game environments | Agentic coding and decision-making | Gert Labs rankings |
| OSWorld-Verified OSWorld-Verified | 75 | #18 | 369 real-world computer tasks (361 when eight Google Drive tasks are excluded) | Multi-step desktop and cross-application workflows | OSWorld |
| CyberGym CyberGym | 79.0 | #19 | 1,507 vulnerability analysis instances | Real-world cybersecurity | CyberGym: Evaluating AI Agents' Real-World Cybersecurity Capabilities at Scale |
| ExploitGym ExploitGym | 6.0 | #21 | 898 exploitation tasks | Advanced cybersecurity exploitation | ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks? |
| JobBench JobBench | 38.9 | #21 | 130 tasks across 35 occupations | Professional multi-source workflows | JobBench: Aligning Agent Work With Human Will |
| MCP Atlas MCP Atlas | 70.6 | #34 | Tool-integrated agent tasks | Advanced tool use | Introducing GPT-5.4 mini and nano |
| Toolathlon Toolathlon | 54.6 | #6 | Multi-tool workflows | Advanced tool use | Introducing GPT-5.4 mini and nano |
| τ²-bench results τ²-Bench Tool-Agent-User Evaluation | 98.9 | #2 | Airline, retail, and telecom customer-service task sets | Dual-control customer-service workflows | τ²-Bench: Evaluating Conversational Agents in a Dual-Control Environment |
| DeepSearchQA DeepSearchQA | 73.6 | #21 | Agentic browsing and list-answer questions | Agentic web research | Muse Spark Eval Methodology |
| Claw-Eval Claw-Eval | 60.3 | #27 | 300 tasks, 2,159 rubrics | Real-world general, multi-turn, and native multimodal agent execution | Claw-Eval: Towards Trustworthy Evaluation of Autonomous Agents |
| ResearchClawBench ResearchClawBench | 15.3 | #13 | 40 tasks across 10 scientific domains | Scientific research re-discovery | ResearchClawBench: A Benchmark for End-to-End Autonomous Scientific Research |
| ApprenticeBench ApprenticeBench: end-to-end computer use, continual learning, and long-horizon agency on a real accounts-payable job | 11 | #14 | 100 vendor bills processed in sequence inside a simulated construction company | Long-horizon computer use with offline and online continual learning | ApprenticeBench: a step change in AI's job readiness |
| MazeBench MazeBench: visual spatial reasoning in a 3D open world | 0 | #29 | 100 hidden gems across 256 rooms in one open world | Long-horizon spatial planning, with single puzzles that can take over one hundred moves | MazeBench Results |
| LifeVal LifeVal: everyday life tasks | 73.0 | #6 | 169 everyday life tasks (caregiving, health, education, household administration, personal finance, shopping, government services) | Agentic personal-assistant work | LifeVal |
| #66 |
| Website generation comparisons |
| Design and website generation |
| OpenRouter Grok 4.3 benchmarks |
| OfficeQA Pro OfficeQA Pro | 53.2 | #14 | Document and spreadsheet tasks | Enterprise grounded reasoning | OfficeQA Pro: An Enterprise Benchmark for End-to-End Grounded Reasoning |
| MMMU-Pro w/ Python MMMU-Pro with Python | 82.1 | #5 | Multimodal academic reasoning | Frontier multimodal | Introducing GPT-5.4 mini and nano |
| ERQA ERQA | 65.4 | #12 | Evidence-based visual QA | Grounded multimodal reasoning | Qwen3.6 launch benchmarks |
| ScreenSpot Pro ScreenSpot Pro | 85.4 | #3 | 1,581 grounding instructions | Professional GUI grounding | ScreenSpot-Pro: GUI Grounding for Professional High-Resolution Computer Use |
| MedXpertQA (MM) MedXpertQA Multimodal | 77.1 | #4 | 2,000 multimodal medical questions | Clinical multimodal reasoning | Muse Spark Eval Methodology |
| ZeroBench ZeroBench | 41.0 | #2 | 100 visual reasoning questions | Tool-augmented visual reasoning | Muse Spark Eval Methodology |
| SimpleVQA SimpleVQA | 61.1 | #12 | Visual QA tasks | General visual understanding | GLM-5V-Turbo |
| CharXiv CharXiv Reasoning | 82.8 | #25 | Scientific chart reasoning | Scientific visualization reasoning | CharXiv: Charting Gaps in Realistic Chart Understanding in Multimodal LLMs |
| ParseBench-1 ParseBench-1: document parsing for AI agents (LlamaIndex leaderboard, first release) | 61.72 | #27 | 2,078 human-verified enterprise pages from insurance, finance and government documents | No method is strong on all five dimensions; charts and visual grounding separate the field | ParseBench: A Document Parsing Benchmark for AI Agents |
| #26 |
| 60 photos of three IKEA builds, some with deliberate assembly mistakes |
| Spatial reasoning from technical diagrams to real photos |
| Furniture Assembly Benchmark (Epoch AI) |
| EBR-bench Earthborne Rangers (EBR-bench) | 25.4 | #21 | Ten playthroughs of an Earthborne Rangers campaign segment per sample (eight learning, two scored) | Learning an unfamiliar game over multi-hour playthroughs | Earthborne Rangers (EBR-bench), Epoch AI |
| MirrorCode MirrorCode (Epoch AI and METR) | 23.3 | #35 | 30 reimplementation tasks (15 programs in two languages), three attempts each | Days-long software engineering without source code | MirrorCode (Epoch AI) |
| Clocktower Bench Clocktower Bench (Blood on the Clocktower) | 1276 | #9 | 386 rated head-to-head games of Blood on the Clocktower (10 to 12 players) | Social deduction, deception and multi-agent reasoning | Clocktower Bench leaderboard |
| ExploitGym Hardened ExploitGym Hardened (stronger exploit mitigations) | 0.3 | #5 | ExploitGym tasks with stronger exploit mitigations | Exploitation against mitigations | ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks? |
| Joke Arena Joke Arena | 1444 | #23 | 100 joke prompts in 10 categories, rated by human votes | Humour writing | BenchLM |
| Game Arena Go Game Arena Go | 856 | #23 | Head-to-head Go games on a 13 x 13 board | Strategic planning and rule-following | BenchLM |
| BinaryAudit BinaryAudit (backdoor detection in compiled binaries) | 17.5 | #13 | 19 backdoor and timebomb detection tasks on stripped binaries, three runs each (57 runs); 6 tooling and 8 clean-binary tasks reported alongside | Reverse engineering without source code or symbols | We hid backdoors in ~40 MB binaries and asked AI + Ghidra to find them (Quesma) |
| NYT Connections NYT Connections LLM Benchmark (extended version) | 91.3 | #20 | 940 NYT Connections puzzles, each with up to four added trick words | Wordplay, word knowledge and decoys that fit almost-but-not-quite | NYT Connections LLM Benchmark (Lech Mazur, GitHub) |