AA-3D: Interactive 3D Experience Benchmark
AA-3D
Clasificación de AA-3D
Rango | Modelo | Elo | |
|---|---|---|---|
| 1 | 1-2 |  Claude Sonnet 5.5 (Max, Default Fallback) | 1199±33Intervalo del 95%: 1166 a 1231 |
| 2 | 1-3 | Claude Opus 5.5 (Max, Default Fallback) | 1196±33Intervalo del 95%: 1164 a 1229 |
| 3 | 3-5 |  GPT-6 Astra (Max) | 1164±32Intervalo del 95%: 1132 a 1196 |
| 4 | 3-6 |  Gemini 4 Argon (High) | 1153±33Intervalo del 95%: 1120 a 1186 |
| 5 | 3-8 | Claude Fable 5.1 (Max, Default Fallback) | 1135±32Intervalo del 95%: 1103 a 1167 |
| 6 | 4-8 | GPT-6 Sol (Max) | 1121±32Intervalo del 95%: 1089 a 1153 |
| 7 | 5-8 | GPT-6.1 Sol (Max) | 1119±32Intervalo del 95%: 1088 a 1151 |
| 8 | 5-8 |  Kimi K3.1 Pro | 1110±33Intervalo del 95%: 1078 a 1143 |
| 9 | 9 |  Muse Spark 1.3 (Max) | 1033±31Intervalo del 95%: 1003 a 1064 |
| 10 | 10 | GPT-6 Luna (Max) | 1000AnclaFijado en 1000 como punto de referencia de la escala Elo de esta clasificación. |
| 11 | 10-14 | Claude Haiku 5.5 (Max) | 998±31Intervalo del 95%: 967 a 1028 |
| 12 | 10-15 |  Grok 4.7 (Xhigh) | 995±31Intervalo del 95%: 964 a 1025 |
| 13 | 10-15 |  MiMo-V2.6-Pro | 983±31Intervalo del 95%: 952 a 1014 |
| 14 | 10-15 | Kimi K3 (Max) | 982±31Intervalo del 95%: 951 a 1013 |
| 15 | 12-17 | Gemini 3.8 Flash (High) | 966±31Intervalo del 95%: 935 a 997 |
| 16 | 15-17 |  Qwen3.8 Max (0902) | 945±31Intervalo del 95%: 914 a 976 |
| 17 | 15-17 |  GLM-5.3-Flash | 937±31Intervalo del 95%: 906 a 967 |
| 18 | 18-19 |  Mistral Large 4 Preview | 870±31Intervalo del 95%: 839 a 901 |
| 19 | 18-19 |  Step 5 Preview | 854±32Intervalo del 95%: 821 a 886 |
| 20 | 20 | GLM-5.3 (Max) | 783±34Intervalo del 95%: 749 a 817 |
| 21 | 21 |  MiniMax-M3 | 740±33Intervalo del 95%: 707 a 773 |
| 22 | 22-23 |  Inkling (Xhigh) | 583±40Intervalo del 95%: 543 a 623 |
| 23 | 22-23 |  Nemotron 3 Ultra 550B A55B (Reasoning) | 573±43Intervalo del 95%: 529 a 616 |
AA-3D Elo
AA-3D Elo
AA-3D: Elo vs. costo por tarea
Elige un dominio, una tarea y un modelo para cargar la escena que creó ese modelo.
Tarea
Modelo
Create an interactive adult man performing weightlifting exercises, with views that expose his muscles, skeleton and organs in anatomically accurate form.
23 de 23 modelos tienen una escena para esta tarea en la votación a ciegas.
Elo vs. Intelligence Index
Elo de AA-3D vs. Intelligence Index
Tasa de victorias por dominio
Cada celda muestra la proporción de votos que un modelo ganó en un dominio de tareas, a partir de los votos de la clasificación actual. La vista equilibrada por rival promedia las tasas de victorias del modelo contra cada rival del dominio, para que cada rival cuente por igual.
Tasa de victorias de AA-3D por dominio
Costo
AA-3D: costo por tarea
Uso de tokens
AA-3D: tokens de salida por tarea
Turnos por escena
Turnos por escena de AA-3D
Uso de herramientas
Llamadas a herramientas por escena de AA-3D
Cómo funciona AA-3D
Cada modelo trabaja como agente en un sandbox sin conexión. Recibe una descripción escrita de una escena interactiva y un contrato de entrega que comparten todas las tareas. El modelo escribe código, renderiza la escena, revisa sus propias capturas de pantalla y hace correcciones. Tiene a su disposición Blender, física Rapier, Python, un navegador headless y una biblioteca de recursos CC0. Una ejecución se detiene a las 3 horas o tras 250 turnos. El resultado es una página index.html que ejecuta la escena en un navegador.
Dentro de una ejecución del agente
Estas repeticiones usan los datos del lanzamiento de AA-3D. Elige una tarea para reproducir la ejecución de cada modelo con un reloj compartido. Pasa el cursor sobre una fila para ver cada acción y haz clic en ella para ver en qué se fue el tiempo.
Mira cómo se crea una escena
Puntuación vs. fecha de lanzamiento
AA-3D: puntuación vs. fecha de lanzamiento
Preguntas frecuentes
AA-3D v1.0 mide qué tan bien los modelos de IA crean experiencias 3D interactivas. Cada modelo recibe una descripción escrita de la tarea y crea una sola página web con three.js en un sandbox sin conexión. El benchmark tiene 55 tareas en 7 dominios. Lee la metodología de AA-3D
Los modelos se clasifican con un sistema de calificación Elo derivado de comparaciones a ciegas realizadas por nuestro panel humano reclutado. Las personas del panel interactúan con dos escenas creadas para la misma tarea y eligen la que prefieren, sin saber qué modelo creó cada una. Un Elo más alto indica que un modelo se prefiere con más frecuencia. Prueba la Votación de AA-3D
Explorar evaluaciones
Artificial Analysis Intelligence Index v5.0A composite benchmark aggregating nine challenging evaluations to provide a holistic measure of AI capabilities across knowledge work, coding, science, and general knowledge.
AA-Briefcase v1.1: Agentic Knowledge Work BenchmarkA private evaluation developed by Artificial Analysis for frontier agentic capability in long-horizon knowledge work, testing agents on realistic business workflows that require deliverables such as spreadsheets, presentations, and memos.
AA-3D: Interactive 3D Experience BenchmarkA benchmark developed by Artificial Analysis for how well AI models build interactive 3D experiences. Models work as agents in an offline sandbox via Stirrup to turn a written brief into a three.js scene, with Elo ratings derived from blind pairwise votes.
APEX-Agents-AA BenchmarkArtificial Analysis' implementation of the APEX-Agents benchmark, testing AI agents on long-horizon, cross-application tasks in professional-services environments with realistic application tooling.
AA-AnalystAgent BenchmarkArtificial Analysis' data analysis benchmark, testing AI agents on their ability to work with spreadsheets and documents to answer quantitative questions a Business Analyst or Data Analyst would face day-to-day.
AutomationBench-AA: Agentic SaaS Workflow BenchmarkA benchmark measuring agentic task completion across simulated SaaS application environments, scoring the share of each task's objectives completed without guardrail violations.
Harvey LAB-AA v1.1 BenchmarkArtificial Analysis' implementation of Harvey's Legal Agent Benchmark (LAB), testing AI agents on real-world legal work from Harvey's dataset of 120 private tasks spanning 24 legal practice areas. v1.1, built in collaboration with Harvey, checks every deliverable against the task's source documents for hallucinations, so work that passes every rubric criterion but contains a material hallucination receives no credit. A panel of three judges grades each deliverable criterion-by-criterion.
GDP.pdf BenchmarkArtificial Analysis' implementation of Surge AI's GDP.pdf benchmark, testing whether language models can reason over long, real-world professional documents and satisfy detailed task-specific criteria.
EnterpriseOps-Gym-AA BenchmarkArtificial Analysis' independent implementation of ServiceNow's EnterpriseOps-Gym, an agentic benchmark testing whether LLM agents can complete stateful, multi-step enterprise workflows across eight business domains via live tool use, graded on the final state of the underlying databases.
Terminal-Bench 4.0 BenchmarkA harder 66-task benchmark of complex terminal work across software, machine learning, science, operations, security, hardware, and media, with recalibrated compute and time allowances and improved instructions, environments, and verifiers.
Terminal-Bench-Science 0.1 BenchmarkA 70-task benchmark of research workflows authored and reviewed by domain experts across the life, physical, earth, mathematical, and engineering sciences, each completed in a terminal and checked by its own set of tests.
FrontierCode 2.0 BenchmarkArtificial Analysis' implementation of Cognition's FrontierCode 2.0, a private agentic coding benchmark of 132 tasks in real open-source repositories, graded on whether a maintainer would merge the agent's change.
HLE-Diamond BenchmarkA refined and updated 1,000-question set of Humanity's Last Exam, split evenly between reasoning and knowledge questions, with 272 questions that include an image.
AA-PersonalityBench: AI Personality & Communication Style BenchmarkMeasures how models communicate rather than whether they are correct: 30 judged personality traits rolled up into 8 indexes, plus 21 measured communication characteristics.
AA-Omniscience v1.1: Knowledge and Hallucination BenchmarkA benchmark measuring factual recall and hallucination across various economically relevant domains.
The most challenging 198 questions from GPQA, where PhD experts achieve 65% accuracy but skilled non-experts only reach 34% despite web access.
ITBench-AA BenchmarkArtificial Analysis' implementation of IBM's ITBench benchmark, testing AI agents on Kubernetes incident root-cause analysis from offline incident snapshots. The agent inspects alerts, events, traces, and topology and identifies the contributing-factor entities (deployments, pods, namespaces, network policies, etc.) responsible for the failure.
MMMU-Pro BenchmarkAn enhanced MMMU benchmark that eliminates shortcuts and guessing strategies to more rigorously test multimodal models across 30 academic disciplines.
IFBench BenchmarkA benchmark evaluating precise instruction-following generalization on 58 diverse, verifiable out-of-domain constraints that test models' ability to follow specific output requirements.
MLCR-AA: Medical Long Context Reasoning BenchmarkAn open benchmark from Wisedocs measuring how well models reason over long, fragmented medical records, performing the multi-document synthesis claims professionals rely on when reviewing insurance and healthcare cases.
Code2Video-AA BenchmarkArtificial Analysis' implementation of HeyGen's Code2Video Bench, evaluating how well models create short product and brand videos from 114 design briefs using HyperFrames HTML.
HealthBench BenchmarkOpenAI's HealthBench: 5,000 health conversations graded against physician-written rubrics, from emergency referrals to global health.
CritPt v2 BenchmarkCritPt v2 is the revised version of CritPt, a benchmark designed to test LLMs on research-level physics reasoning tasks, featuring 70 composite research challenges. Models solve problems with a code-execution tool in an offline Linux sandbox.
GDPval-AA v2.1 BenchmarkGDPval-AA v2.1 is Artificial Analysis' evaluation framework for OpenAI's GDPval dataset. It tests AI models on real-world tasks across 44 occupations and 9 major industries. Models are given shell access and web browsing capabilities in an agentic loop via Stirrup to solve tasks, with Elo ratings derived from blind pairwise comparisons.
𝜏³-Banking BenchmarkA fintech customer-support benchmark from the 𝜏-Knowledge framework that tests whether agents can navigate a large unstructured knowledge base and execute multi-step tool calls to resolve realistic banking workflows.
Terminal-Bench 2.1 BenchmarkA verified refresh of Terminal-Bench 2.0: 89 curated tasks across software engineering, system administration, data processing, model training, and security, with environment and instruction fixes so scores reflect agent capability rather than environment gaps.
FrontierCode 1.1 BenchmarkArtificial Analysis' implementation of Cognition's FrontierCode 1.1, a private agentic coding benchmark that tests models on completing coding tasks in a way that would actually get merged by maintainers of real-world repositories.
AA-LCR v1.1: Long Context Reasoning BenchmarkA challenging benchmark measuring language models' ability to extract, reason about, and synthesize information from long-form documents ranging from 10k to 100k tokens (measured using the cl100k_base tokenizer).
SciCode BenchmarkA scientist-curated coding benchmark featuring 288 test set subproblems from 80 laboratory problems across 16 scientific disciplines.
Humanity's Last Exam BenchmarkA frontier-level benchmark with 2,500 expert-vetted questions across mathematics, sciences, and humanities, designed to be the final closed-ended academic evaluation.
CritPt BenchmarkA benchmark designed to test LLMs on research-level physics reasoning tasks, featuring 71 composite research challenges.
Terminal-Bench Hard BenchmarkAn agentic benchmark evaluating AI capabilities in terminal environments through software engineering, system administration, and data processing tasks.
𝜏²-Bench Telecom BenchmarkA dual-control conversational AI benchmark simulating technical support scenarios where both agent and user must coordinate actions to resolve telecom service issues.
MMLU-Pro BenchmarkAn enhanced version of MMLU with 12,000 graduate-level questions across 14 subject areas, featuring ten answer options and deeper reasoning requirements.
LiveCodeBench BenchmarkA contamination-free coding benchmark that continuously harvests fresh competitive programming problems from LeetCode, AtCoder, and CodeForces, evaluating code generation, self-repair, and execution.
MATH-500 BenchmarkA 500-problem subset from the MATH dataset, featuring competition-level mathematics across six domains including algebra, geometry, and number theory.
AIME 2025 BenchmarkAll 30 problems from the 2025 American Invitational Mathematics Examination, testing olympiad-level mathematical reasoning with integer answers from 000-999.
Global-MMLU-Lite BenchmarkA lightweight, multilingual version of MMLU, designed to evaluate knowledge and reasoning skills across a diverse range of languages and cultural contexts.