Artificial Analysis
Artificial Analysis is an independent AI model evaluation and comparison platform focused on conducting multi-dimensional benchmarking of large language models and other AI models. Metrics covered include model capability (Intelligence Index), inference speed, pricing, and availability. Operating on principles of independence and objectivity, the platform aggregates results from multiple standardized tests to provide cross-model comparative references for researchers, developers, and enterprises.
Core Facts
Artificial Analysis is a third-party evaluation platform that visualizes and compares model intelligence and operational cost on the same coordinate system.
Artificial Analysis gave Argon an intelligence index score of 53, on par with GPT-6 Astra.
Credibility is higher only when multiple independent evaluation sources reach consistent conclusions, and multiple evaluation sources should be cross-validated.
Timeline (last 90 days)
On the new version of the Artificial Analysis intelligence index, Carbon ranks first with 59 points, ahead of Claude Opus 5.5's 58 points and GPT-6 Astra's 57 points.
On the fifth edition of the Comprehensive Intelligence leaderboard, Gemini 4 Carbon scored 59 points, ranking first and overtaking Claude Opus 5.5, 9 points higher than Argon.
Artificial Analysis estimates that the comprehensive performance gap between top Chinese and US models has narrowed to 3.3 months, while APAC AI data shows that the United States monopolizes 78% of the world's actual effective computing power.
Gemini 4 Carbon took first place on the text leaderboard in its first Arena test with a score of 1528 points.
NVIDIA recently released the Nemotron 3.5 Transcribe speech transcription model.
Artificial Analysis Intelligence Index V5 added coding and scientific research evaluations, of which 60% use private data or confidential answers.
In the Intelligence Index V5 released by Artificial Analysis, Cloud Opus 5.5 ranks first with 58 points, and DeepSeek V4.1 Flash ties for first among open-weight models with 32 points.
According to Artificial Analysis data, DeepSeek V4.1 Flash can reach 1286 tokens per second via Lithos AI, and Base 10 provides an inference speed of 845 tokens per second.
Gemini 4 Carbon scored 59 on the Artificial Analysis Intelligence Index.
The industry has formed a stricter reference framework: prioritize the results of independent evaluation organizations (such as LMSYS Chatbot Arena and Artificial Analysis), then check whether the evaluation code and runtime environment are published, and only lastly refer to vendors' self-reported leaderboards.
40 more timeline events
All Facts (20)
Artificial Analysis is a third-party evaluation platform that visualizes and compares model intelligence and operational cost on the same coordinate system.
85%VerifiedArtificial Analysis gave Argon an intelligence index score of 53, on par with GPT-6 Astra.
80%VerifiedCredibility is higher only when multiple independent evaluation sources reach consistent conclusions, and multiple evaluation sources should be cross-validated.
75%VerifiedArtificial Analysis is an independent AI model evaluation organization founded in 2023 and headquartered in London.
75%VerifiedThe Artificial Analysis Intelligence Index provides a unified intelligence score by integrating multiple evaluations including reasoning, coding, mathematics, and knowledge Q&A.
75%VerifiedOpus 5.5 scored 66 on the Terminal agent benchmark, higher than Sonnet's 55 and Astra's 57.9.
70%VerifiedArtificial Analysis是一个独立的AI模型基准测试平台，综合编程能力、数学推理、语言理解、知识问答等多个维度评估模型表现
70%VerifiedVals AI's independent leaderboard places Gemini 4 Argon first among 41 models (68.9%).
65%VerifiedThird-party platforms such as Artificial Analysis normalize scores from multiple sub-benchmarks and aggregate them into a single quality index, enabling different models to be compared on the same scale.
65%VerifiedArtificial Analysis evaluation shows DeepSeek V4.1 Flash scored 40 on the Intelligence Index, surpassing V4 Pro to become DeepSeek's new flagship, featuring native multimodal input, 8B activated parameters on input and 16B on output, released under the MIT license.
65%VerifiedQwen3.8-Flash-Next has approximately 6B activated parameters.
65%VerifiedArtificial Analysis Intelligence Index整合了MMLU、HumanEval、MATH、GPQA等多个基准测试的加权结果
65%VerifiedElevenLabs has launched a new v4 voice synthesis model.
65%VerifiedArtificial Analysis released version v4.2 of the Intelligence Index.
65%VerifiedMoonshot AI has released a massive open-source model called Kimi K3 with 2.8 trillion parameters, using a MoE architecture.
65%Partially VerifiedIn coding agent scenarios, Claude Code on Opus 5.5 scored 66, Wonder scored 63, and only Wonder Full can reach 68.
70%UnverifiedThe Artificial Analysis Index (AA) is published by the independent evaluation organization Artificial Analysis, aggregating benchmarks such as MMLU, GPQA, HumanEval, and MATH to calculate a composite score from 0 to 100.
70%UnverifiedGLM-5.3-Flash achieved a score of 57 on the Artificial Analysis composite intelligence index evaluation, on par with Claude Opus 4.8.
70%UnverifiedGDPval（General Domain Performance Validation）是一个综合性大语言模型评估基准，涵盖推理、知识问答、代码生成、多语言理解等维度
70%Partially VerifiedKimi K3.1 scored 55 on the Artificial Analysis composite benchmark.
65%