- 图表数据
2391 张图 · 73,917 个数据点 · 12 家实验室
每张图，都是数据。
实验室发布的图表背后的数字，以及每个数字的来源。研究它们、重画它们，或做你自己的。标有“相似重绘”的图表有一个得分 7 分以上（满分 10）的相似重绘。
图表2391
数据点73,917
发布方数字31全部或部分
从图片估算2360
实验室12
已通过的相似重绘95相似度 7 分以上（满分 10）
OpenAI 211
全部模型从图片估算相似重绘 7.4TerminalBench 2.1OpenAI · 柱状图 · 9 个数据点从图片估算相似重绘 8.1Illustrative global managed assetsOpenAI · 分组柱状图 · 22 个数据点从图片估算相似重绘 8.1Illustrative global managed assetsOpenAI · 分组柱状图 · 22 个数据点部分为发布方的数字相似重绘 8.3Accuracy / Percentile (%)deepseek · 跑分柱状图 · 30 个数据点从图片估算Model Benchmark ComparisonOpenAI · 表格图 · 42 个数据点从图片估算GPT4 Turbo Benchmark ComparisonOpenAI · 折线图 · 12 个数据点从图片估算Grok-4.20-Beta1 ranks #1 in Search Arenaarena · 表格图 · 40 个数据点从图片估算Benchmark ComparisonOpenAI · 表格图 · 27 个数据点从图片估算Artificial Analysis Intelligence Index v4.1.1artificial-analysis · 柱状图 · 29 个数据点从图片估算Artificial Analysis Intelligence Index v4.1OpenAI · 散点图 · 14 个数据点从图片估算相似重绘 8.1Artificial Analysis Intelligence Index v4.1artificial-analysis · 分组柱状图 · 30 个数据点从图片估算相似重绘 8.9Artificial Analysis Intelligence Index v4.1artificial-analysis · 柱状图 · 44 个数据点从图片估算Agent Performance vs Humans with Time Limits (CI 95%)OpenAI · 柱状图 · 6 个数据点从图片估算Neurons spike rate, Paw movement, Licking, Mouth movement, Run speed, Pupil diameter, global astrocytic signalOpenAI · 折线图 · 40 个数据点从图片估算相似重绘 9.2Illustrative global managed assets have more than doubled since 2015OpenAI · 分组柱状图 · 22 个数据点从图片估算ExploitBenchOpenAI · 折线图 · 22 个数据点从图片估算FrontierScience-Research accuracy across modelsOpenAI · 柱状图 · 8 个数据点从图片估算相似重绘 5.7Change in combined output tokens by department since November 2025OpenAI · 折线图 · 20 个数据点从图片估算OpenAI o3-pro: comparative evaluations with human testersOpenAI · 柱状图 · 5 个数据点从图片估算Model Performance ComparisonOpenAI · 柱状图 · 4 个数据点从图片估算AI Model Benchmark ComparisonOpenAI · 表格图 · 30 个数据点从图片估算AI Model Benchmark ComparisonOpenAI · 表格图 · 31 个数据点从图片估算Visual-6 Luna (Max) · RefCOCOgOpenAI · 柱状图 · 2 个数据点从图片估算Artificial Analysis Coding IndexOpenAI · 折线图 · 34 个数据点从图片估算FrontierScience-Olympiad accuracy across modelsOpenAI · 柱状图 · 8 个数据点从图片估算AIME 2025 Competition mathOpenAI · 跑分柱状图 · 9 个数据点从图片估算HealthBench ProfessionalOpenAI · 折线图 · 54 个数据点从图片估算BrowseCompOpenAI · 柱状图 · 6 个数据点从图片估算Internal instruction following evalOpenAI · 柱状图 · 2 个数据点从图片估算相似重绘 8.1CyberGymOpenAI · 柱状图 · 6 个数据点从图片估算SWE-bench Verified Software engineeringOpenAI · 折线图 · 7 个数据点从图片估算GPT-4 Pricing Over TimeOpenAI · 折线图 · 4 个数据点从图片估算GDPval Knowledge work tasksOpenAI · 堆叠柱状图 · 6 个数据点从图片估算AccuracyOpenAI · 跑分柱状图 · 20 个数据点从图片估算GPT-4o mini is more than 60% cheaper than GPT-3.5 TurboOpenAI · 折线图 · 10 个数据点从图片估算GPT-Rosalind improves core biology and chemistry benchmarksOpenAI · 跑分柱状图 · 20 个数据点从图片估算Recent released open weight models compare to frontier modelsOpenAI · 折线图 · 60 个数据点从图片估算SWE-Bench Verified Software EngineeringOpenAI · 柱状图 · 4 个数据点从图片估算Non-developerOpenAI · 折线图 · 33 个数据点从图片估算Up to 1.5MOpenAI · 柱状图 · 4 个数据点从图片估算Occupation vs. work tokens done with CodexOpenAI · 表格图 · 25 个数据点从图片估算Internal hallucination with noise evalOpenAI · 柱状图 · 4 个数据点从图片估算Common VoiceOpenAI · 柱状图 · 2 个数据点从图片估算DeveloperOpenAI · 折线图 · 33 个数据点从图片估算OpenAI o3: 80% price reduction over previous o3OpenAI · 分组柱状图 · 6 个数据点从图片估算Artificial Analysis Coding Agent Index v1.1OpenAI · 散点图 · 31 个数据点从图片估算GPT-5.6 Sol on the ARC-AGI-3 Public SetOpenAI · 折线图 · 10 个数据点从图片估算Coding deception (lower is better)OpenAI · 柱状图 · 5 个数据点从图片估算GPQA DiamondOpenAI · 分组柱状图 · 7 个数据点从图片估算Net Improvement vs. median cost per session (7-day window)arena · 散点图 · 15 个数据点从图片估算Image Edit Arena-2.5: Ranked #1arena · 柱状图 · 17 个数据点从图片估算MATH Level 5 accuracy for o3 and o4-mini compared to leading modelsepoch-ai · 柱状图 · 8 个数据点从图片估算GPT-Live-1 Performance BenchOpenAI · 跑分柱状图 · 6 个数据点从图片估算Code GPT-6 Arena: WebDev (Max): Rank #24arena · 柱状图 · 24 个数据点从图片估算GPT-6Sol(Max) Pareto Frontier for Code Arenaarena · 折线图 · 9 个数据点从图片估算SWE-Lancer IC SWEOpenAI · 柱状图 · 2 个数据点从图片估算OpenAI o3 ranks #5 in WebDev Arena!arena · 表格图 · 33 个数据点从图片估算Artificial Analysis Hallucination Rateartificial-analysis · 柱状图 · 40 个数据点从图片估算Reviews with high scores by evaluation criteriaOpenAI · 跑分柱状图 · 20 个数据点从图片估算LifeSciBench Scores by Scientific WorkflowOpenAI · 跑分柱状图 · 18 个数据点从图片估算Share of tasks Agent solved on a given attempt (pass@1) vs pass@5 (at least once) vs pass^5, on all 5 attempts (pass^5) - Top 10 models by pass^5 - Higher is betterartificial-analysis · 跑分柱状图 · 36 个数据点从图片估算SWE-bench VerifiedOpenAI · 柱状图 · 2 个数据点从图片估算SWE-Bench ProOpenAI · 柱状图 · 3 个数据点从图片估算Common VoiceOpenAI · 柱状图 · 2 个数据点从图片估算Artificial Analysis Speech to Speech Index by Componentartificial-analysis · 跑分柱状图 · 100 个数据点从图片估算OpenAI o3-pro: 87% price reduction compared to o1-proOpenAI · 分组柱状图 · 8 个数据点从图片估算Artificial Analysis Coding Agent Indexartificial-analysis · 柱状图 · 15 个数据点从图片估算GPT-5.5 Instant: better at visual reasoningOpenAI · 跑分柱状图 · 4 个数据点从图片估算Artificial Conversational DynamicsOpenAI · 柱状图 · 3 个数据点从图片估算Competition Math (AIME 2024)arena · 柱状图 · 6 个数据点从图片估算Intelligence vs. Price: Non-Reasoning Modelsartificial-analysis · 散点图 · 14 个数据点从图片估算o3 #1 for Style Control, Coding, Hard o3 & o4 mini #1 for Matharena · 表格图 · 130 个数据点从图片估算GPT-5.6 Sol and Luna (xHigh)arena · 散点图 · 6 个数据点从图片估算GPT-5.5 Instant: a step up in STEMOpenAI · 跑分柱状图 · 4 个数据点从图片估算Nano Banana 2.1: Gap to Use Case Frontier (AA-Image-T2I v2.0)artificial-analysis · 表格图 · 30 个数据点从图片估算Terminal-Bench 4.0OpenAI · 折线图 · 15 个数据点从图片估算Common VoiceOpenAI · 柱状图 · 4 个数据点从图片估算GPT-6.1-Sol (Max) Pareto Frontier for Code Arenaarena · 折线图 · 6 个数据点从图片估算AutomationBenchOpenAI · 折线图 · 32 个数据点从图片估算DeepSWEOpenAI · 折线图 · 84 个数据点从图片估算Fraction of Model A Wins for All Non-tied A vs. B Battlesarena · 表格图 · 81 个数据点从图片估算SWE-Bench Pro (Public)OpenAI · 折线图 · 15 个数据点从图片估算Context Aware ASR BenchmarkOpenAI · 跑分柱状图 · 4 个数据点从图片估算Average time to First audio (seconds) on Big Bench Audio dataset - Lower is betterartificial-analysis · 柱状图 · 26 个数据点从图片估算FrontierMath accuracyepoch-ai · 散点图 · 6 个数据点从图片估算Output Tokens used Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 42 个数据点从图片估算Artificial Analysis Quality Indexartificial-analysis · 柱状图 · 18 个数据点从图片估算FrontierMath private accuracy for OpenAI's o3 and o4-mini by reasoning settingepoch-ai · 柱状图 · 6 个数据点从图片估算Artificial Analysisartificial-analysis · 柱状图 · 36 个数据点从图片估算DeepSWEOpenAI · 折线图 · 16 个数据点从图片估算GPT Image 2.5 Sunburst (max): Gap to Use Case Frontier (Image Editing)artificial-analysis · 表格图 · 20 个数据点从图片估算Image-to-WebDev: Ranked #1arena · 跑分柱状图 · 15 个数据点从图片估算Code Arenaarena · 柱状图 · 13 个数据点从图片估算SWE-bench Verified accuracyOpenAI · 柱状图 · 7 个数据点从图片估算Artificial Analysis Image Arena Leaderboard (Text to Image)artificial-analysis · 表格图 · 10 个数据点从图片估算Cost (USD) to Run all Evaluations in the Artificial Intelligence Indexartificial-analysis · 跑分柱状图 · 31 个数据点从图片估算Output Speedartificial-analysis · 柱状图 · 25 个数据点从图片估算Chat CompletionsOpenAI · 折线图 · 9 个数据点从图片估算GPT-5.1 ranks #3 & #4 in Visionarena · 表格图 · 45 个数据点从图片估算Artificial Analysis Intelligence Index v3.0artificial-analysis · 柱状图 · 34 个数据点从图片估算Factuality Leaderboardarena · 柱状图 · 20 个数据点从图片估算Training GPT-4o on 9x more compute would have yielded o1-high's performanceepoch-ai · 散点图 · 6 个数据点从图片估算Code GPT-6 Astra (Max): Ranked #1arena · 柱状图 · 24 个数据点从图片估算o1-high vastly outperforms GPT-4o on GPQA diamond and MATH level 5epoch-ai · 跑分柱状图 · 4 个数据点从图片估算Multi-Image Edit: Arena Ranked #1 & #2arena · 柱状图 · 19 个数据点从图片估算Text-to-Image Arena Ranks #2arena · 柱状图 · 17 个数据点从图片估算Internal OpenAI Instructions following eval accuracy (hard subset)OpenAI · 柱状图 · 8 个数据点从图片估算Real World Audio RecordingOpenAI · 柱状图 · 4 个数据点从图片估算Artificial Analysis vs. Output Tokens Used in Artificial Analysis Intelligence Indexartificial-analysis · 散点图 · 24 个数据点从图片估算OSWorld-VerifiedOpenAI · 柱状图 · 3 个数据点从图片估算OpenAI o3 #2 GPT-4.1 & o4-mini in the top 10!arena · 表格图 · 14 个数据点从图片估算Factual error rate on difficult prompts (lower is better)OpenAI · 折线图 · 21 个数据点从图片估算GPT Image 2.5 Sunburst (max): Gap to Editing Action Frontierartificial-analysis · 表格图 · 14 个数据点从图片估算Context Aware ASR BenchmarkOpenAI · 跑分柱状图 · 6 个数据点从图片估算OTIS Mock AIME accuracy for OpenAI's o3 and o4-mini compared to Gemini 2.5 Pro and leading modelsepoch-ai · 柱状图 · 8 个数据点从图片估算Artificial Analysis Speech to Speech Indexartificial-analysis · 跑分柱状图 · 82 个数据点从图片估算OpenAI GPT-to-Image-1 ranks #1 in Text-to-Image Arena!arena · 表格图 · 27 个数据点从图片估算Reduction in AI scheming rates after deliberative alignmentOpenAI · 分组柱状图 · 4 个数据点从图片估算OpenAI o3-pro: pass@1 evaluation benchmarksOpenAI · 跑分柱状图 · 9 个数据点从图片估算o1 was a major algorithmic advanceepoch-ai · 分组柱状图 · 11 个数据点从图片估算Output Tokens per Second; Higher is betterartificial-analysis · 柱状图 · 18 个数据点从图片估算SEC-Bench Pro (Multi-Agent)OpenAI · 折线图 · 17 个数据点从图片估算Professional Capture-the-Flag ChallengesOpenAI · 散点图 · 6 个数据点从图片估算r²-Bench Telecom Benchmark Leaderboard: Resultsartificial-analysis · 柱状图 · 32 个数据点从图片估算Artificial Analysis Cyber Index: Score vs. Cost per Taskartificial-analysis · 散点图 · 19 个数据点从图片估算GPT-Live-1 Performance Bench v3 Tool callingOpenAI · 分组柱状图 · 3 个数据点从图片估算Generation Time (Text to Image)artificial-analysis · 柱状图 · 3 个数据点从图片估算Cost per Hour of Input Audio (Big Bench Audio subset)artificial-analysis · 柱状图 · 18 个数据点从图片估算Artificial Analysis Speech to Speech Index vs. time to first audio - Only models with all data available shown - Higher index and lower latency are betterartificial-analysis · 散点图 · 13 个数据点从图片估算Speech Reasoning (Big Bench Audio)artificial-analysis · 柱状图 · 24 个数据点从图片估算Artificial Analysis Intelligence Index vs. Output Tokens per Intelligence Index Taskartificial-analysis · 散点图 · 30 个数据点从图片估算OpenAI's Sora debuts on the Video Arena Leaderboard at #11arena · 表格图 · 14 个数据点从图片估算Sora 2 Pro tie ranks #1arena · 表格图 · 12 个数据点从图片估算Warmth Scoreartificial-analysis · 柱状图 · 5 个数据点从图片估算Code GPT-6 Sol: WebDev Ranked #4arena · 柱状图 · 20 个数据点从图片估算Arena Preference Elo vs. Task Success Rateartificial-analysis · 跑分柱状图 · 20 个数据点从图片估算SWE-Bench Pro (public)OpenAI · 折线图 · 12 个数据点从图片估算Meta Llama 3 Instructartificial-analysis · 堆叠柱状图 · 12 个数据点从图片估算Output tokens to run a set of aggregated benchmarkssemianalysis · 堆叠柱状图 · 32 个数据点从图片估算Score by Benchmarkartificial-analysis · 跑分柱状图 · 57 个数据点从图片估算Internal moderation eval (multi-policy accuracy)OpenAI · 柱状图 · 6 个数据点从图片估算Arena Rankings Updatearena · 表格图 · 24 个数据点从图片估算GPT Image 2.5 Flare (max): Gap to Capability Frontierartificial-analysis · 表格图 · 27 个数据点从图片估算Open Weights: Frontier Indexartificial-analysis · 折线图 · 30 个数据点从图片估算Quality vs Price Frontier (Image Editing)artificial-analysis · 散点图 · 24 个数据点从图片估算Artificial Analysis Coding Agent Index v1.1artificial-analysis · 表格图 · 16 个数据点从图片估算Average OpenAI tool calls per model response across generationssemianalysis · 柱状图 · 8 个数据点从图片估算Intelligence Index vs. Output Tokens per Intelligence Index Taskartificial-analysis · 散点图 · 26 个数据点从图片估算Artificial Intelligence Indexartificial-analysis · 柱状图 · 13 个数据点从图片估算Harvey LAB-AA v1.1: Hallucination Judge Model Comparisonartificial-analysis · 表格图 · 66 个数据点从图片估算Competition mathOpenAI · 跑分柱状图 · 10 个数据点从图片估算Intelligence vs. Total Parametersartificial-analysis · 散点图 · 14 个数据点从图片估算torchtitan gptosssemianalysis · 跑分柱状图 · 5 个数据点从图片估算Big Bench Audio: IntelligenceOpenAI · 柱状图 · 2 个数据点从图片估算Text-to-Image Arenaarena · 柱状图 · 15 个数据点从图片估算Gemini also ranks #1 on Text-to-Image Arenaarena · 表格图 · 12 个数据点从图片估算Qwen3.5-397B-A17B #17 Overall in Code Arenaarena · 表格图 · 27 个数据点部分为发布方的数字Input Priceartificial-analysis · 柱状图 · 21 个数据点从图片估算#1 - Grok Imagine Videoarena · 表格图 · 8 个数据点从图片估算Artificial Intelligence Analysis Intelligence Index: Non-Reasoning Modelsartificial-analysis · 柱状图 · 16 个数据点从图片估算Harvey LAB-AA v1.1: Hallucination-Gated All-Pass Rate vs. Cost per Taskartificial-analysis · 散点图 · 18 个数据点从图片估算Terminal Bench 2.0OpenAI · 柱状图 · 2 个数据点从图片估算Energy consumption per ChatGPT query is small compared to everyday electricity useepoch-ai · 柱状图 · 7 个数据点从图片估算Artificial Analysis Cyber Index: Score vs. Cost per Taskartificial-analysis · 散点图 · 22 个数据点从图片估算Pareto Frontier for Agent Arenaarena · 散点图 · 9 个数据点从图片估算OpenAI MRCRv2, 8 needlesOpenAI · 折线图 · 10 个数据点从图片估算相似重绘 8.3Text-to-Image Arena Top 20arena · 柱状图 · 25 个数据点从图片估算Agent 5.6: Sol, Terra & Lunaarena · 柱状图 · 30 个数据点从图片估算Real World Audio RecordingOpenAI · 柱状图 · 2 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 柱状图 · 21 个数据点从图片估算Conversational Dynamics (Full Duplex Bench v1.5 - Higher is better)artificial-analysis · 分组柱状图 · 16 个数据点从图片估算Users long-running queries, run remotely, are disproportionately at the end of a user's last request of the dayOpenAI · 分组柱状图 · 15 个数据点从图片估算CritPt Benchmark Leaderboard: Token Usageartificial-analysis · 跑分柱状图 · 60 个数据点从图片估算Advanced Cybersecurity Completion RateOpenAI · 柱状图 · 4 个数据点从图片估算Time to First Audio (seconds) on Big Bench Audio dataset - Lower is betterartificial-analysis · 柱状图 · 24 个数据点从图片估算Image Arena Quality ELOartificial-analysis · 柱状图 · 17 个数据点从图片估算Tau2-bench TelecomOpenAI · 柱状图 · 5 个数据点从图片估算Model Performance Comparisonarena · 表格图 · 79 个数据点从图片估算OSWorld 2.0, offline setOpenAI · 折线图 · 15 个数据点从图片估算Agent Arenaarena · 表格图 · 6 个数据点从图片估算Model Evaluation ScoresOpenAI · 跑分柱状图 · 40 个数据点从图片估算GPT-6Sol(Max) Pareto Frontier for Agent Arenaarena · 散点图 · 11 个数据点从图片估算Artificial Analysis Intelligence Index v2.2artificial-analysis · 柱状图 · 19 个数据点从图片估算GDP.pdfOpenAI · 折线图 · 24 个数据点从图片估算Artificial Analysis Intelligence Index v4.1artificial-analysis · 表格图 · 69 个数据点从图片估算Cost per complete Hour of Input Audio (Big Bench Audio subset)artificial-analysis · 柱状图 · 19 个数据点从图片估算Terminal-Bench 2.0OpenAI · 柱状图 · 3 个数据点从图片估算Terminal-Bench 2.0OpenAI · 柱状图 · 3 个数据点从图片估算AA-Briefcase v1.1: Scoreartificial-analysis · 分组柱状图 · 21 个数据点从图片估算CritPt Benchmark Leaderboard: Resultsartificial-analysis · 分组柱状图 · 18 个数据点从图片估算ELO score in Artificial Analysis Image Arena (relative metric of image generation quality), Price: USD per 1000 image generationsartificial-analysis · 散点图 · 20 个数据点从图片估算Tau3 (Voice) IntelligenceOpenAI · 跑分柱状图 · 9 个数据点从图片估算Pareto Frontier for Image-to-WebDevarena · 散点图 · 10 个数据点从图片估算Grok Imagine Image 2.0: Quality vs Speed Frontier (Text to Image)artificial-analysis · 散点图 · 19 个数据点从图片估算Goblins increased in GPT-5.4 especially for the Nerdy personalityOpenAI · 柱状图 · 8 个数据点从图片估算Codeforces Competition codeOpenAI · 柱状图 · 7 个数据点从图片估算Agents' Last ExamOpenAI · 折线图 · 24 个数据点从图片估算Artificial Analysis Text to Video Leaderboardsartificial-analysis · 表格图 · 12 个数据点从图片估算Artificial Analysis Cyber Index: Score by Benchmarkartificial-analysis · 跑分柱状图 · 51 个数据点从图片估算Agent Arenaarena · 柱状图 · 25 个数据点从图片估算Audio MultiChallenge: Instruction FollowingOpenAI · 柱状图 · 2 个数据点从图片估算SWE-Bench Verified (n=500)OpenAI · 折线图 · 7 个数据点从图片估算GPT-5.2-chat-latest lands top 5 in Text Arenaarena · 表格图 · 10 个数据点从图片估算Text-to-Image-2.5 Arena: Ranked #1arena · 柱状图 · 18 个数据点从图片估算Humanity's Last ExamOpenAI · 柱状图 · 7 个数据点从图片估算GPT-5 improves on o4-mini on SWE-bench Verified, but still scores below Claude 4.1epoch-ai · 柱状图 · 6 个数据点从图片估算Artificial Analysis Intelligence Index vs. Cost per Intelligence Index Taskartificial-analysis · 散点图 · 42 个数据点从图片估算FrontierMath private accuracy for OpenAI's o3 and o4-mini compared to leading modelsepoch-ai · 柱状图 · 14 个数据点从图片估算Computer-use safety stress test (lower is better)OpenAI · 柱状图 · 4 个数据点从图片估算SMB % of Agentic TokensOpenAI · 折线图 · 8 个数据点从图片估算Overall model performanceOpenAI · 柱状图 · 16 个数据点
Anthropic 239
全部模型从图片估算Claude 3.5 Sonnet benchmarksAnthropic · 表格图 · 34 个数据点从图片估算Software engineeringAnthropic · 柱状图 · 2 个数据点从图片估算Novel problem-solving by costAnthropic · 散点图 · 3 个数据点从图片估算相似重绘 7.4Code Arena: WebDevarena · 柱状图 · 15 个数据点从图片估算Model benchmarksAnthropic · 表格图 · 31 个数据点从图片估算相似重绘 6.0Opus 4.8Anthropic · 表格图 · 28 个数据点从图片估算Frontend Code Kimi-K3: Ranked #1arena · 柱状图 · 23 个数据点从图片估算Claude 4 benchmarksAnthropic · 表格图 · 65 个数据点从图片估算相似重绘 6.0Artificial Analysis Intelligence Index v4.1artificial-analysis · 分组柱状图 · 30 个数据点从图片估算相似重绘 7.1Artificial Analysis Intelligence Index v4.1artificial-analysis · 柱状图 · 44 个数据点从图片估算Claude Mythos 5 and Fable 5Anthropic · 表格图 · 67 个数据点从图片估算Opus 5Anthropic · 表格图 · 51 个数据点从图片估算相似重绘 2.4Firefox Security Vulnerabilities by MonthAnthropic · 堆叠柱状图 · 56 个数据点从图片估算AA-Briefcase Eloartificial-analysis · 柱状图 · 35 个数据点从图片估算相似重绘 7.7Long context retrievalAnthropic · 折线图 · 16 个数据点从图片估算相似重绘 6.6Agentic business Automation workflows by effortAnthropic · 折线图 · 36 个数据点从图片估算Agentic codingAnthropic · 折线图 · 31 个数据点从图片估算相似重绘 7.8Model Performance ComparisonAnthropic · 表格图 · 46 个数据点从图片估算相似重绘 7.7Agentic computer useAnthropic · 散点图 · 18 个数据点从图片估算相似重绘 8.3Agentic computer use performance by effort levelAnthropic · 折线图 · 19 个数据点从图片估算相似重绘 9.3SoftwareAnthropic · 柱状图 · 6 个数据点从图片估算Agentic terminalAnthropic · 折线图 · 21 个数据点从图片估算CursorBench3.2.0 Agentic codingAnthropic · 折线图 · 10 个数据点从图片估算相似重绘 7.4SoftwareAnthropic · 柱状图 · 6 个数据点从图片估算Agentic codingAnthropic · 折线图 · 21 个数据点从图片估算Sonnet 5Anthropic · 表格图 · 18 个数据点部分为发布方的数字Claude Haiku 5.5Anthropic · 表格图 · 42 个数据点从图片估算Humanity's Last Exam Multidisciplinary reasoningAnthropic · 折线图 · 36 个数据点从图片估算Claude 3.7 Sonnet benchmarksAnthropic · 表格图 · 56 个数据点从图片估算相似重绘 8.0Business workflowsAnthropic · 折线图 · 21 个数据点从图片估算Where a researcher went wrong, could Claude have done better?Anthropic · 跑分柱状图 · 20 个数据点部分为发布方的数字Claude Fable 5.1Anthropic · 表格图 · 53 个数据点从图片估算Model Benchmark ComparisonAnthropic · 表格图 · 77 个数据点从图片估算Indexed cost of Fable usageAnthropic · 跑分柱状图 · 12 个数据点从图片估算相似重绘 2.7MultidisciplinaryAnthropic · 折线图 · 30 个数据点从图片估算相似重绘 4.0Real-world knowledgeAnthropic · 折线图 · 25 个数据点从图片估算Opus 4.7Anthropic · 表格图 · 64 个数据点从图片估算AI usage by job typeAnthropic · 分组柱状图 · 42 个数据点从图片估算相似重绘 8.0Sonnet 4.6Anthropic · 表格图 · 83 个数据点从图片估算Finding vulnerabilities and exploits in open-source codeAnthropic · 分组柱状图 · 12 个数据点部分为发布方的数字Claude Opus 5.5Anthropic · 表格图 · 39 个数据点从图片估算Computer useAnthropic · 折线图 · 10 个数据点从图片估算Terminal-Bench-Science 0.1 Agentic scientific researchAnthropic · 折线图 · 10 个数据点从图片估算Agentic codingAnthropic · 折线图 · 22 个数据点从图片估算Terminal-Bench 4.0 Agentic terminal codingAnthropic · 折线图 · 15 个数据点从图片估算SoftwareAnthropic · 柱状图 · 6 个数据点从图片估算Agentic terminalAnthropic · 折线图 · 27 个数据点从图片估算Misaligned behaviorAnthropic · 柱状图 · 4 个数据点从图片估算Knowledgework by effort levelAnthropic · 折线图 · 20 个数据点从图片估算GPT-6.1 Sol and Claude Opus 5.5: API value by plansemianalysis · 分组柱状图 · 6 个数据点从图片估算Agent Arenaarena · 柱状图 · 14 个数据点从图片估算Pareto Frontier for Text Arenaarena · 散点图 · 9 个数据点从图片估算Agentic coding runs on cachesemianalysis · 表格图 · 18 个数据点从图片估算AA-Briefcase Presentation Elo vs. Rubric Score (%)artificial-analysis · 散点图 · 38 个数据点从图片估算Sycophancy by model (levels 1-2)Anthropic · 分组柱状图 · 8 个数据点从图片估算Intelligence Analysis Index vs. Cost per Taskartificial-analysis · 散点图 · 25 个数据点从图片估算Arena Score Top15arena · 柱状图 · 13 个数据点从图片估算GDPval-AA Leaderboardartificial-analysis · 柱状图 · 27 个数据点从图片估算Multilingual Index (English)artificial-analysis · 柱状图 · 12 个数据点从图片估算CritPt Benchmark Developed by Argonne and UIUC: Scoreartificial-analysis · 柱状图 · 35 个数据点从图片估算Output Tokens per Secondartificial-analysis · 柱状图 · 17 个数据点从图片估算Cost per Token type Intelligence Index Taskartificial-analysis · 表格图 · 30 个数据点从图片估算How Claude Models Writearena · 分组柱状图 · 24 个数据点从图片估算AutomationBenchopenai · 折线图 · 27 个数据点从图片估算Claude Opus 5.5 leads on software engineering, GPT-6 Astra on mathepoch-ai · 表格图 · 11 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 柱状图 · 25 个数据点从图片估算Elo rating-AA for performance on real-world work tasks - Anchored to a human baseline of 1,000 - Higher is betterartificial-analysis · 柱状图 · 36 个数据点从图片估算GDPpdf: All-passartificial-analysis · 分组柱状图 · 26 个数据点从图片估算Expert GPT-5.5: Top #5arena · 柱状图 · 17 个数据点从图片估算GDPval scores for agentic performance on real-world work tasks using web and shell access via Stirrup, an open-source harness developed by Artificial Analysisartificial-analysis · 柱状图 · 33 个数据点从图片估算Claude 3.5 Sonnet (new) benchmarksAnthropic · 表格图 · 26 个数据点从图片估算Claude Opus 5.5 (High) & Pareto in Agent Arenaarena · 柱状图 · 15 个数据点从图片估算Intelligence Evaluationsartificial-analysis · 分组柱状图 · 11 个数据点从图片估算How long does Claude Code work before stopping?Anthropic · 折线图 · 69 个数据点从图片估算GPQA Diamond accuracy for Claude Sonnet 4 and Claude Opus 4 compared to other modelsepoch-ai · 柱状图 · 9 个数据点从图片估算AA-Share of tasks Agent solved on all 5 of pass^5 Leaderboardartificial-analysis · 柱状图 · 36 个数据点从图片估算Terminal-Bench v4.0: Scoreartificial-analysis · 跑分柱状图 · 36 个数据点从图片估算Artificial Analysis Intelligence Index vs. Output Tokens per Intelligence Index Taskartificial-analysis · 散点图 · 26 个数据点从图片估算Software engineeringAnthropic · 跑分柱状图 · 10 个数据点从图片估算AA-Briefcase Rubric Score (%)artificial-analysis · 分组柱状图 · 39 个数据点从图片估算Code Arena: Frontendarena · 柱状图 · 25 个数据点从图片估算Coding, Hard Prompts, Writing & Longer Query improved significantly over 3.7arena · 表格图 · 119 个数据点从图片估算How Claude Models Writearena · 分组柱状图 · 28 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 柱状图 · 39 个数据点从图片估算Total Coding Market ARRsemianalysis · 折线图 · 132 个数据点从图片估算Intelligence vs. Time per Intelligence Index Taskartificial-analysis · 散点图 · 70 个数据点从图片估算Harvey LAB-AA: All-pass rate vs. Average cost per Task (USD)artificial-analysis · 散点图 · 29 个数据点从图片估算Kimi K3 is the highest-ECI open model to date, CIs are 90% confidence intervalsepoch-ai · 跑分柱状图 · 11 个数据点从图片估算Mean Turns per Taskartificial-analysis · 柱状图 · 36 个数据点从图片估算Intelligence Analysis vs. Output Tokensartificial-analysis · 散点图 · 31 个数据点从图片估算Artificial Analysis Coding Agent Indexartificial-analysis · 分组柱状图 · 13 个数据点从图片估算GPQA Diamond (Scientific Reasoning)artificial-analysis · 柱状图 · 15 个数据点从图片估算Estimating Claude's productivity impactsAnthropic · 分组柱状图 · 8 个数据点从图片估算Agent Arena: Codearena · 柱状图 · 16 个数据点从图片估算Artificial Analysis Intelligence Index v4.1artificial-analysis · 跑分柱状图 · 61 个数据点从图片估算Artificial Analysis Intelligence Index incorporates 7 evaluationsartificial-analysis · 柱状图 · 19 个数据点从图片估算AA-Omniscence Index: Scoreartificial-analysis · 跑分柱状图 · 79 个数据点从图片估算Artificial Analysis Intelligence Index v3.0artificial-analysis · 柱状图 · 31 个数据点从图片估算Harvey Share of LAB tasks-AA: All-pass criterion Rateartificial-analysis · 柱状图 · 37 个数据点从图片估算Multilingual Index (French)artificial-analysis · 柱状图 · 12 个数据点从图片估算What a dollar of ChatGPT and Claude buys todaysemianalysis · 散点图 · 11 个数据点从图片估算How Claude Models Writearena · 分组柱状图 · 29 个数据点从图片估算SWE-bench Verified accuracy for Claude Sonnet 4 and Claude Opus 4 compared to other modelsepoch-ai · 柱状图 · 18 个数据点从图片估算Net Production Models PerformanceAnthropic · 跑分柱状图 · 25 个数据点从图片估算SWE-Atlas@1 on SWE-Atlas-QnA - Higher is betterartificial-analysis · 柱状图 · 19 个数据点从图片估算Claude 3.7 Sonnet gain was a major algorithmic advanceepoch-ai · 跑分柱状图 · 12 个数据点从图片估算Agent GPT-5.5 (xHigh) Leaderboard: Ranked #2arena · 跑分柱状图 · 20 个数据点从图片估算How the work changed over seven monthsAnthropic · 折线图 · 63 个数据点从图片估算Code Arena: WebDevarena · 柱状图 · 15 个数据点从图片估算Time per Taskartificial-analysis · 分组柱状图 · 38 个数据点从图片估算CritPt Benchmark Leaderboard: Resultsartificial-analysis · 分组柱状图 · 32 个数据点从图片估算What kinds of failures are flagged?arena · 表格图 · 115 个数据点从图片估算Artificial Analysis Intelligence Index v2.1artificial-analysis · 表格图 · 36 个数据点从图片估算Claude Opus 4.7 ranks #1 in Code Arenaarena · 表格图 · 20 个数据点从图片估算AA-Briefcase Eloartificial-analysis · 柱状图 · 23 个数据点从图片估算LiveCodeBench (Coding): Non-Reasoning Modelsartificial-analysis · 柱状图 · 12 个数据点从图片估算SuccessAnthropic · 分组柱状图 · 31 个数据点从图片估算Terminal-Bench-Science 0.1 score vs. average Cost per Task (USD) - Lower is betterartificial-analysis · 散点图 · 134 个数据点从图片估算EnterpriseOps-Gym-AA score vs. average cost per task (Cost USD) - Lower is betterartificial-analysis · 散点图 · 28 个数据点从图片估算AA-Briefcase Elo vs. Cost per Taskartificial-analysis · 散点图 · 28 个数据点从图片估算Artificial Analysis Coding Agent Index v1.4artificial-analysis · 柱状图 · 21 个数据点从图片估算Output Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 36 个数据点从图片估算Output Tokens used Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 36 个数据点从图片估算Artificial Analysis Intelligence Index v4.1artificial-analysis · 柱状图 · 42 个数据点从图片估算Claude Opus 4.7 Thinking ranks #1 in Vision Arenaarena · 表格图 · 20 个数据点从图片估算SWE-bench Verified accuracy for Claude Sonnet 4 and Claude Opus 4 compared to previous Claude modelsepoch-ai · 柱状图 · 10 个数据点从图片估算Claude Opus 4 tied #1 in WebDev Arenaarena · 表格图 · 33 个数据点从图片估算Software engineeringAnthropic · 柱状图 · 6 个数据点从图片估算FrontierCode 1.1: Score vs. Cost per Taskartificial-analysis · 散点图 · 29 个数据点从图片估算Claude models playing Pokémon*Anthropic · 折线图 · 44 个数据点从图片估算Anthropic in the Text Arenaarena · 表格图 · 60 个数据点从图片估算Code Arena: Frontendarena · 柱状图 · 17 个数据点从图片估算Average cost per Taskartificial-analysis · 柱状图 · 19 个数据点从图片估算Fraction of examples with faithful CoTAnthropic · 跑分柱状图 · 4 个数据点从图片估算How Claude Models Writearena · 分组柱状图 · 24 个数据点从图片估算Claude Opus 4.1 ranks #2 on WebDev (tied with Gemini 2.5 Pro)arena · 表格图 · 11 个数据点从图片估算Mythos Preview outpaced the trend in cyber frontier models (best scores at release)epoch-ai · 折线图 · 28 个数据点从图片估算AA-AnalystAgent:pass^5 vs Cost per Taskartificial-analysis · 散点图 · 30 个数据点从图片估算Arena Opus 4.7 vs Opus 4.6 Text Category Rankingsarena · 折线图 · 28 个数据点从图片估算Image-to-WebDev Arena Top 25: Overallarena · 柱状图 · 25 个数据点从图片估算Intelligence Index vs. Cost per Intelligence Index Taskartificial-analysis · 散点图 · 12 个数据点从图片估算Claude Opus 4.7 ranks #1 in Document Arenaarena · 表格图 · 20 个数据点从图片估算GPT-5 is #1 on WebDev Arenaarena · 表格图 · 8 个数据点从图片估算Agent Arena Leaderboardarena · 柱状图 · 25 个数据点从图片估算How Claude Models Writearena · 分组柱状图 · 29 个数据点从图片估算Claude Code interruption rate by experienceAnthropic · 折线图 · 16 个数据点从图片估算Multilingual Index (Spanish)artificial-analysis · 柱状图 · 13 个数据点从图片估算MMLU (Quality - Claude 3 comparison)artificial-analysis · 柱状图 · 10 个数据点从图片估算Task Complexity, Max. Consecutive Tool Calls, and Number of Human TurnsAnthropic · 跑分柱状图 · 6 个数据点从图片估算WebDev Arena Leaderboardarena · 表格图 · 16 个数据点从图片估算Harvey LAB-AA all-pass: All-pass Rate vs. Output Tokens per Taskartificial-analysis · 散点图 · 29 个数据点从图片估算WebDev Arena Leaderboardarena · 表格图 · 24 个数据点从图片估算Intelligence vs. Cost to Run Artificial Analysis Intelligence Indexartificial-analysis · 散点图 · 17 个数据点从图片估算Artificial Analysis Coding Index (Non-Reasoning Models)artificial-analysis · 柱状图 · 12 个数据点从图片估算Score by Benchmarkartificial-analysis · 跑分柱状图 · 66 个数据点从图片估算Terminal pass@1 on Terminal-Bench v2artificial-analysis · 柱状图 · 21 个数据点从图片估算Artificial Analysis Intelligence Index v4.1artificial-analysis · 分组柱状图 · 27 个数据点从图片估算MultidisciplinaryAnthropic · 折线图 · 10 个数据点从图片估算Claude Sonnet 4.5 #4 for WebDev Arenaarena · 表格图 · 36 个数据点从图片估算SciCode: Scoreartificial-analysis · 跑分柱状图 · 64 个数据点从图片估算Plan usage limitsAnthropic · 柱状图 · 3 个数据点从图片估算Claude Opus 5.5 (High) Pareto Frontier for Agent Arenaarena · 折线图 · 10 个数据点从图片估算MMLU vs. Priceartificial-analysis · 散点图 · 10 个数据点从图片估算How Claude Models Writearena · 分组柱状图 · 24 个数据点从图片估算Token Usage, cache, and Output tokens per Taskartificial-analysis · 堆叠柱状图 · 75 个数据点从图片估算Artificial Analysis Coding Indexartificial-analysis · 柱状图 · 20 个数据点从图片估算Output Tokens per Intelligence Index Taskartificial-analysis · 散点图 · 40 个数据点从图片估算How Claude Models Writearena · 表格图 · 27 个数据点从图片估算DeepSWE Benchmark Scoreartificial-analysis · 柱状图 · 15 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 柱状图 · 20 个数据点从图片估算DeepSeek V3.1 excels at Math, Creative Writing and Longer Query (Top 3)arena · 表格图 · 135 个数据点从图片估算Iterative improvement on a frontier-scale modelAnthropic · 折线图 · 59 个数据点从图片估算Outputartificial-analysis · 柱状图 · 41 个数据点从图片估算AA-Briefcase Elo vs. Cost per Taskartificial-analysis · 散点图 · 63 个数据点从图片估算AA-Briefcase Eloartificial-analysis · 分组柱状图 · 35 个数据点从图片估算AutomationBench-AA: Scoreartificial-analysis · 分组柱状图 · 35 个数据点从图片估算Artificial Analysis Coding Agent Indexartificial-analysis · 柱状图 · 10 个数据点从图片估算AA-Briefcase Eloartificial-analysis · 柱状图 · 40 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 柱状图 · 20 个数据点从图片估算Code GPT-6 Astra: WebDev Win Rates in Battlesarena · 分组柱状图 · 15 个数据点从图片估算Artificial Analysis Legal Index - Weighted average wall clock time per Taskartificial-analysis · 散点图 · 22 个数据点从图片估算Intelligence Analysis vs. Cost per Intelligence Index Taskartificial-analysis · 散点图 · 80 个数据点从图片估算Text #5 Labs: Best Models Category Rankingsarena · 表格图 · 40 个数据点从图片估算Optimize throughput with prompt cachingAnthropic · 堆叠柱状图 · 4 个数据点从图片估算Code Arena Top 15: Agentic Codingarena · 柱状图 · 14 个数据点从图片估算Output Tokens per Intelligence Index Taskartificial-analysis · 跑分柱状图 · 70 个数据点从图片估算Claude Code+ Commits Over Timesemianalysis · 折线图 · 12 个数据点从图片估算Text Arenaarena · 柱状图 · 17 个数据点从图片估算Test cases passed for Opus 4.6 reimplementing pkl in Rustepoch-ai · 折线图 · 105 个数据点从图片估算Artificial Analysis Intelligence Index v4.3artificial-analysis · 分组柱状图 · 28 个数据点从图片估算Latency vs. Throughput: Provider Chunk & Model combinationsartificial-analysis · 散点图 · 8 个数据点从图片估算Model Benchmark Comparisonopenai · 表格图 · 43 个数据点从图片估算How Claude Models Write Language Markersarena · 分组柱状图 · 28 个数据点从图片估算Newer models score higher for same token budgetepoch-ai · 折线图 · 12 个数据点从图片估算Frontier Analysis Language Intelligence Index v4.3.2 Intelligence By Country, Over Timeartificial-analysis · 折线图 · 57 个数据点从图片估算Task duration where we predict 50% chance of AI successmetr · 柱状图 · 6 个数据点从图片估算Claude 3.5 Haiku benchmarksAnthropic · 表格图 · 26 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 34 个数据点从图片估算Artificial Analysis Indexesartificial-analysis · 跑分柱状图 · 80 个数据点从图片估算AA-Briefcase Eloartificial-analysis · 柱状图 · 33 个数据点从图片估算Code Arena Top 15: Agentic Codingarena · 柱状图 · 14 个数据点从图片估算Claude tool calls per model responsesemianalysis · 柱状图 · 5 个数据点从图片估算GDPval-AA v2 Leaderboardartificial-analysis · 柱状图 · 25 个数据点从图片估算Artificial Analysis Intelligence Index v4.1artificial-analysis · 柱状图 · 26 个数据点从图片估算Terminal-Bench-Science 0.1 score vs. average Cost per Task (USD) - Lower is betterartificial-analysis · 散点图 · 12 个数据点从图片估算Usage Share Across Occupational CategoriesAnthropic · 分组柱状图 · 15 个数据点从图片估算Harvey LAB-AA: All-pass Rate vs. Time per Taskartificial-analysis · 散点图 · 30 个数据点从图片估算How Claude Models Write Language Markersarena · 分组柱状图 · 24 个数据点从图片估算相似重绘 3.0Artificial Analysis Intelligence Indexartificial-analysis · 分组柱状图 · 18 个数据点从图片估算AA-Briefcase Eloartificial-analysis · 柱状图 · 29 个数据点从图片估算AA-Briefcase v1.1artificial-analysis · 分组柱状图 · 53 个数据点从图片估算MMLU (Quality benchmark comparison)artificial-analysis · 柱状图 · 11 个数据点从图片估算Output Tokens per Intelligence Index Taskartificial-analysis · 表格图 · 22 个数据点从图片估算Real-world knowledgeAnthropic · 折线图 · 16 个数据点从图片估算Fraction of examples with faithful CoTAnthropic · 分组柱状图 · 5 个数据点从图片估算Clarification and interruption rates by goal complexityAnthropic · 分组柱状图 · 8 个数据点从图片估算Agent Arena Claude Sonnet 5arena · 表格图 · 6 个数据点从图片估算Image-to-WebDev Arena Ranked #2arena · 柱状图 · 15 个数据点从图片估算Intelligence Evaluationsartificial-analysis · 跑分柱状图 · 88 个数据点从图片估算Multilingual Index (German)artificial-analysis · 柱状图 · 12 个数据点从图片估算Pareto Frontier for Agent Arenaarena · 散点图 · 13 个数据点从图片估算Pricing per 1M Tokensartificial-analysis · 分组柱状图 · 24 个数据点从图片估算Code Arena: WebDev GPT-6 Astra vs Claude Fable 5.1arena · 表格图 · 16 个数据点从图片估算Text Arena Claude Sonnet 5 vs 4.6 Category Rankingsarena · 表格图 · 30 个数据点从图片估算Harvey LAB-AA Benchmark Leaderboard: All-pass Rate vs. Average Turns per Taskartificial-analysis · 散点图 · 40 个数据点从图片估算Terminal-Bench Hard Artificial Analysis Leaderboard: Resultsartificial-analysis · 柱状图 · 31 个数据点从图片估算AI performance solving issues from 12 open-source Python repositoriesepoch-ai · 散点图 · 22 个数据点从图片估算Claude Opus 4.7 scores near the frontierepoch-ai · 柱状图 · 5 个数据点从图片估算How Claude models differAnthropic · 分组柱状图 · 12 个数据点从图片估算Terminal-Bench 4.0: Scoreartificial-analysis · 跑分柱状图 · 36 个数据点从图片估算Gemini 2.5 Pro also jumps to #2 on WebDev Arena!arena · 表格图 · 30 个数据点从图片估算Pareto Frontier for Code Arenaarena · 折线图 · 7 个数据点从图片估算Code Arena: WebDevarena · 表格图 · 14 个数据点从图片估算Output Tokens per Intelligence Index Taskartificial-analysis · 散点图 · 24 个数据点从图片估算Claude 3.7 Sonnet Thinking shows strength across a variety of categoriesarena · 表格图 · 150 个数据点从图片估算Misaligned cyberattacks in pursuit of reward...Anthropic · 分组柱状图 · 8 个数据点从图片估算Misaligned actions in pursuit of rewardAnthropic · 分组柱状图 · 8 个数据点从图片估算Simulation Inspired by Hugging Face IncidentAnthropic · 分组柱状图 · 31 个数据点从图片估算US GDP (trillions of dollars a year, in 2025 prices)Anthropic · 折线图 · 30 个数据点
Google 179
全部模型从图片估算相似重绘 8.3Gemini 3.8 Flash CyberGoogle · 柱状图 · 3 个数据点从图片估算Gemini 2.5 Flash Native AudioGoogle · 跑分柱状图 · 7 个数据点从图片估算CWE-bench v1 leaderboardGoogle · 柱状图 · 11 个数据点从图片估算Gemini 3 Deep ThinkGoogle · 跑分柱状图 · 18 个数据点从图片估算相似重绘 8.2Speech Reasoning: Big Bench Audioartificial-analysis · 柱状图 · 14 个数据点从图片估算Discovering securityGoogle · 跑分柱状图 · 4 个数据点从图片估算Eva Bench - Experience to accuracyGoogle · 散点图 · 13 个数据点从图片估算Artificial Analysis Intelligence Index v3.0artificial-analysis · 柱状图 · 27 个数据点从图片估算相似重绘 9.2Gemini 4 Argon Benchmark ComparisonGoogle · 表格图 · 88 个数据点从图片估算相似重绘 6.1Code Gemini 4 Argon: WebDev Ranked #8arena · 柱状图 · 39 个数据点从图片估算WebDev Arena Leaderboard 05-06-25 02:00 PTGoogle · 表格图 · 48 个数据点从图片估算Text Arena Gemini 4 Argon: Ranked #1arena · 柱状图 · 25 个数据点从图片估算Gray Swan IPIGoogle · 堆叠柱状图 · 42 个数据点从图片估算相似重绘 7.8Artificial Analysis Intelligence Index vs. Cost to Run Artificial Analysis Intelligence Indexartificial-analysis · 散点图 · 34 个数据点从图片估算Arena Score vs Cost per 1M Output TokensGoogle · 散点图 · 34 个数据点从图片估算Massive Audio Embedding BenchmarkGoogle · 散点图 · 9 个数据点从图片估算相似重绘 7.6Model Performance VS SizeGoogle · 散点图 · 10 个数据点从图片估算相似重绘 7.6Generalization BenchmarkGoogle · 跑分柱状图 · 9 个数据点从图片估算Text to ImageGoogle · 跑分柱状图 · 18 个数据点从图片估算Speed & Cost EfficiencyGoogle · 分组柱状图 · 29 个数据点从图片估算Unlock token efficiency and accuracy gainsGoogle · 跑分柱状图 · 12 个数据点从图片估算Humanity's Last ExamGoogle · 柱状图 · 4 个数据点从图片估算CodeforcesGoogle · 柱状图 · 3 个数据点从图片估算Speech to SpeechGoogle · 柱状图 · 6 个数据点从图片估算Gemini 3.6 FlashGoogle · 跑分柱状图 · 12 个数据点从图片估算Enterprise workflow automationGoogle · 柱状图 · 4 个数据点从图片估算FLEURS (top locales*)Google · 柱状图 · 5 个数据点从图片估算Gemini 2.5 Pro Deep ThinkGoogle · 跑分柱状图 · 12 个数据点从图片估算相似重绘 7.0Speech to Speech IndexGoogle · 柱状图 · 6 个数据点从图片估算Time to solve (minutes)Google · 分组柱状图 · 21 个数据点从图片估算Humanity's Last ExamGoogle · 柱状图 · 4 个数据点从图片估算Long-horizon software engineeringGoogle · 柱状图 · 5 个数据点从图片估算Production FrontierCode 1.1 MainGoogle · 柱状图 · 4 个数据点从图片估算Agentic Performance (t-Voice)Google · 柱状图 · 6 个数据点从图片估算ASIMOV - Safety Instruction FollowingGoogle · 分组柱状图 · 9 个数据点从图片估算Gemini 3.5 Flash-Lite Compared to 3 FlashGoogle · 跑分柱状图 · 4 个数据点从图片估算MMLU Performance vs. Model Sizes (Mix-n-Match Configurations)Google · 散点图 · 16 个数据点从图片估算τ³-Banking LeaderboardGoogle · 柱状图 · 5 个数据点从图片估算Gemini 3.7 FlashGoogle · 柱状图 · 5 个数据点从图片估算Massive Image Embedding Benchmark (Lite)Google · 散点图 · 9 个数据点从图片估算Agentic PerformanceGoogle · 柱状图 · 6 个数据点从图片估算MMMU-ProGoogle · 柱状图 · 4 个数据点从图片估算FLEURS (top locales*)Google · 柱状图 · 6 个数据点从图片估算Massive Text Embedding Benchmark (Code)Google · 散点图 · 9 个数据点从图片估算Gemini 3 Deep ThinkGoogle · 柱状图 · 4 个数据点从图片估算Model Capability ComparisonGoogle · 表格图 · 33 个数据点从图片估算3.7 Flash is the Agentic frontier video understanding on accuracy and costGoogle · 散点图 · 6 个数据点从图片估算Gemini 3.5 Flash-LiteGoogle · 跑分柱状图 · 4 个数据点从图片估算Instruction Following BenchmarkGoogle · 跑分柱状图 · 6 个数据点从图片估算Artificial Intelligence Analysis incorporates Intelligence Indexartificial-analysis · 柱状图 · 22 个数据点从图片估算ITBench-AA: Average precision at full recall vs. Cost per Taskartificial-analysis · 散点图 · 38 个数据点从图片估算Nano Banana 2.1: Gap to Capability Frontier (AA-Image-Editing v2.0)artificial-analysis · 表格图 · 21 个数据点从图片估算Priceartificial-analysis · 柱状图 · 29 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 35 个数据点从图片估算Cost (USD) to run all evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 29 个数据点从图片估算ELO scores from blind preference votes in our Image Arena. See the full leaderboard here.artificial-analysis · 柱状图 · 17 个数据点从图片估算Gemini 4 Argon Pareto Frontier for Text Arenaarena · 散点图 · 10 个数据点从图片估算Gemini 4 Argon Pareto Frontier for Text Arenaarena · 散点图 · 10 个数据点从图片估算Arena Code Agentic WebDev Leaderboardarena · 表格图 · 25 个数据点从图片估算AA-Briefcase Eloartificial-analysis · 分组柱状图 · 26 个数据点从图片估算Output Tokens per Secondartificial-analysis · 柱状图 · 30 个数据点从图片估算Seedream 3.0 ties #4 in Text-to-Image Arena!arena · 表格图 · 27 个数据点从图片估算Code Arena: Frontendarena · 柱状图 · 17 个数据点从图片估算Artificial Analysis Video Arena Leaderboard (Image to Video)artificial-analysis · 表格图 · 18 个数据点从图片估算Chatbot Arena Overview (Task)arena · 表格图 · 117 个数据点从图片估算Pareto Frontier for Agent Arenaarena · 折线图 · 1 个数据点从图片估算Omniscience Accuracy vs. Omniscience Indexartificial-analysis · 散点图 · 33 个数据点从图片估算MMMU-Pro Benchmark Leaderboard: Resultsartificial-analysis · 柱状图 · 24 个数据点从图片估算Artificial Analysis Intelligence Index vs. Cost per Taskartificial-analysis · 散点图 · 34 个数据点从图片估算Nano Banana 2.1: Gap to Capability Frontier (AA-Image-T2I v2.0)artificial-analysis · 表格图 · 27 个数据点从图片估算Single Image Edit Arenaarena · 柱状图 · 14 个数据点从图片估算Intelligence vs. Priceartificial-analysis · 散点图 · 12 个数据点从图片估算GPQA Diamond accuracy for Mistral Small 3 compared to other modelsepoch-ai · 柱状图 · 5 个数据点从图片估算Characters Per secondartificial-analysis · 柱状图 · 44 个数据点从图片估算Text Arenaarena · 柱状图 · 31 个数据点从图片估算Intelligence Parameters vs. Model Parameters: Intelligence Small, Open Weights (Version 2, released Feb '25)artificial-analysis · 散点图 · 11 个数据点从图片估算Intelligence Evaluationsartificial-analysis · 柱状图 · 100 个数据点从图片估算MMMU-Pro Benchmark Leaderboard: Resultsartificial-analysis · 柱状图 · 19 个数据点从图片估算Intelligence vs. Compute Proxyartificial-analysis · 散点图 · 17 个数据点从图片估算GPQA Diamond accuracy for OpenAI's o3 and o4-mini compared to leading modelsepoch-ai · 柱状图 · 7 个数据点从图片估算Pronunciation Robustness by Categoryartificial-analysis · 分组柱状图 · 64 个数据点从图片估算Figure 2: Average Win Rate Against All Other Models (Assuming Uniform Sampling and No Ties)arena · 柱状图 · 8 个数据点从图片估算Image Nano Banana 2.1: Ranked #6arena · 柱状图 · 16 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 分组柱状图 · 26 个数据点从图片估算Speech reasoning: based on the Artificial Analysis Big Bench Audio dataset - Higher is betterartificial-analysis · 柱状图 · 33 个数据点从图片估算Pareto Frontier for Agent Arenaarena · 散点图 · 9 个数据点从图片估算AA-WER Streaming Index (First Partial Transcription) vs. Time to First Partial Transcription After Speech Endartificial-analysis · 散点图 · 28 个数据点从图片估算Output Reasoning Tokens Used to Run Artificial Analysis Intelligence Indexartificial-analysis · 分组柱状图 · 38 个数据点从图片估算Nano Banana 2.1: Quality vs Speed Frontier (AA-Image-T2I v2.0)artificial-analysis · 散点图 · 7 个数据点从图片估算Nano Banana 2.1: Quality vs Price Frontier (AA-Image-T2I v2.0)artificial-analysis · 散点图 · 12 个数据点从图片估算Output Tokens used Run all evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 76 个数据点从图片估算Intelligence Evaluationsartificial-analysis · 分组柱状图 · 44 个数据点从图片估算Artificial Analysisartificial-analysis · 跑分柱状图 · 40 个数据点从图片估算GDPval-AA v2 Leaderboardartificial-analysis · 分组柱状图 · 42 个数据点从图片估算Speech Reasoning: Based on the Artificial Analysis Big Bench Audio dataset; Higher is betterartificial-analysis · 柱状图 · 13 个数据点从图片估算DeepSWE v1.1semianalysis · 散点图 · 10 个数据点从图片估算Artificial Analysis Intelligence Index incorporates 7 evaluations: MMLU-Pro, GPQA Diamond, Humanity's Last Exam, LiveCodeBench, SciCode, AIME, MATH-500artificial-analysis · 柱状图 · 19 个数据点从图片估算Cost per Intelligence Index Taskartificial-analysis · 柱状图 · 45 个数据点从图片估算Artificial Analysis Intelligence Index v4.1artificial-analysis · 分组柱状图 · 40 个数据点从图片估算Output Tokens per Secondartificial-analysis · 柱状图 · 31 个数据点从图片估算Text-to-Image Arena 2.1: Ranked #5arena · 柱状图 · 15 个数据点从图片估算Time per Intelligence Index Taskartificial-analysis · 柱状图 · 25 个数据点从图片估算Image-to-Video Arena 2025arena · 表格图 · 19 个数据点从图片估算Artificial Analysis Omniscience Accuracyartificial-analysis · 柱状图 · 35 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 35 个数据点从图片估算Text Arenaarena · 柱状图 · 26 个数据点从图片估算Text-to-Video Arena 2025arena · 表格图 · 40 个数据点从图片估算Cost per Hour of Input Audio Big Bench (Big Bench Audio subset)artificial-analysis · 柱状图 · 20 个数据点从图片估算GDPval-AA Leaderboardartificial-analysis · 柱状图 · 27 个数据点从图片估算Intelligence vs. Total Parametersartificial-analysis · 散点图 · 49 个数据点从图片估算Output Tokensartificial-analysis · 跑分柱状图 · 78 个数据点从图片估算Provider Elo: Arena Preferenceartificial-analysis · 柱状图 · 38 个数据点从图片估算Intelligence vs. Cost to Run Artificial Analysis Intelligence Indexartificial-analysis · 散点图 · 30 个数据点从图片估算GPT-5 matches the performance of other models on GPQA Diamondepoch-ai · 柱状图 · 8 个数据点从图片估算Artificial Analysis Intelligence Index incorporates 7 evaluations: MMLU-Pro, GPQA Diamond, Humanity's Last Exam, LiveCodeBench, SciCode, AIME, MATH-500artificial-analysis · 柱状图 · 22 个数据点从图片估算Output Tokens per Secondartificial-analysis · 柱状图 · 22 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 17 个数据点从图片估算Output Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 37 个数据点从图片估算Agentic conversation performance (τ-Voice) Average conversation Audio Durationartificial-analysis · 柱状图 · 12 个数据点从图片估算Intelligence Evaluationsartificial-analysis · 跑分柱状图 · 231 个数据点从图片估算Video Arenaarena · 柱状图 · 14 个数据点从图片估算Rank from human voice preference votes on the Artificial Analysis Speech Arena (Controlled voice). Rank is calculated independently for each language. Lower rank is better.artificial-analysis · 表格图 · 110 个数据点从图片估算Artificial Analysis Intelligence Index v4.1artificial-analysis · 柱状图 · 39 个数据点从图片估算Figure 1: Confidence Intervals on Model Imagen Strength (via Bootstrapping)arena · 散点图 · 8 个数据点从图片估算Output Tokens Used to Run Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 65 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 柱状图 · 25 个数据点从图片估算Artificial Analysis Intelligence Index v3.0artificial-analysis · 柱状图 · 24 个数据点从图片估算Pareto Frontierarena · 散点图 · 8 个数据点从图片估算Agent Arenaarena · 柱状图 · 25 个数据点从图片估算Intelligence Index vs. Total Parametersartificial-analysis · 散点图 · 15 个数据点从图片估算Output speed: o200k_base Gemma 4 31B (Reasoning)epoch-ai · 柱状图 · 13 个数据点从图片估算Veo-3.1-1080p ranks #1 & 2 in Video Arenaarena · 表格图 · 30 个数据点从图片估算Intelligence vs. Output Speedartificial-analysis · 散点图 · 33 个数据点从图片估算Key Players in the AI Value Chainartificial-analysis · 表格图 · 96 个数据点从图片估算Agent Arena Gemini 3.8 Flash vs Gemini 3.7 Flasharena · 表格图 · 10 个数据点从图片估算HappyHorse by Alibaba ranks #2 in Text-to-Videoarena · 表格图 · 7 个数据点从图片估算Video Arenaarena · 柱状图 · 13 个数据点从图片估算Pareto Frontier: Text / Overall / Style Control Cost vs Scorearena · 散点图 · 18 个数据点从图片估算Weighted average cost (USD) per Artificial Analysis Intelligence Index: Gemini 3.1 Pro Preview to Gemini 4 Argonartificial-analysis · 跑分柱状图 · 12 个数据点从图片估算Wan2.6-t2v #1 Chinese lab in Video Arenaarena · 表格图 · 30 个数据点从图片估算Arena Gemini 3 Pro vs. 3.1 Pro Category Rankingsarena · 表格图 · 28 个数据点从图片估算Cost to Run Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 57 个数据点从图片估算MATH Level 5 pass@1 for Mistral Small 3 compared to other modelsepoch-ai · 柱状图 · 5 个数据点从图片估算Cost to Run Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 132 个数据点从图片估算Wan 2.5 t2i preview ranks #5 in Text-to-Imagearena · 表格图 · 60 个数据点从图片估算Output Tokens per Secondartificial-analysis · 柱状图 · 33 个数据点从图片估算AA-Briefcase Eloartificial-analysis · 分组柱状图 · 20 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 分组柱状图 · 22 个数据点从图片估算Output Tokens used Run all evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 90 个数据点从图片估算Artificial Analysis Intelligence Index v3.0artificial-analysis · 柱状图 · 41 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 分组柱状图 · 42 个数据点从图片估算GDPval-AA Leaderboardartificial-analysis · 跑分柱状图 · 49 个数据点从图片估算Pareto Frontierarena · 散点图 · 6 个数据点从图片估算Elo scores from blind preference votes in our Image Arenaartificial-analysis · 柱状图 · 20 个数据点从图片估算Agentic Performance (τ-Verb)artificial-analysis · 柱状图 · 26 个数据点从图片估算AA-Briefcase Elo and GDPval-AI v2 Leaderboardartificial-analysis · 跑分柱状图 · 41 个数据点从图片估算Artificial Analysis Word Error Rate (AA-WER) Index by Open Weights / Proprietaryartificial-analysis · 跑分柱状图 · 40 个数据点从图片估算Artificial Analysis Image Arena Leaderboardartificial-analysis · 表格图 · 22 个数据点从图片估算Output Tokens per Intelligence Indexartificial-analysis · 跑分柱状图 · 50 个数据点从图片估算Frontend Gemini-3.6 Code Arena: Rank #12arena · 柱状图 · 26 个数据点从图片估算Text-to-Image Arenaarena · 柱状图 · 15 个数据点从图片估算Arena Preference Elo vs. Cost per Hour of Input Audioartificial-analysis · 散点图 · 21 个数据点从图片估算Text-to-Image Arena Nano Banana 2 Lite Pareto Frontierarena · 折线图 · 7 个数据点从图片估算Veo-3.1-1080p ranks #2 & 5 in Video Arenaarena · 表格图 · 20 个数据点从图片估算Artificial Analysis Speech to Speech Index vs. Cost per Hour of Input Audioartificial-analysis · 散点图 · 12 个数据点从图片估算Output Tokens per Secondartificial-analysis · 柱状图 · 30 个数据点从图片估算Speech to Speech Arena Eloartificial-analysis · 柱状图 · 19 个数据点从图片估算Highlights from Q3, our first-ever $100B revenue quarterGoogle · 表格图 · 10 个数据点从图片估算Q1 Earnings Highlights: AI investments and full stack approach are driving performance across our businessGoogle · 表格图 · 8 个数据点从图片估算Q4 Earnings Highlights: Annual revenues exceeded $400B for the first timeGoogle · 表格图 · 7 个数据点从图片估算Gemini Model Capability ComparisonGoogle · 表格图 · 39 个数据点从图片估算Gemini 3 Deep ThinkGoogle · 跑分柱状图 · 15 个数据点从图片估算Knowledge workGoogle · 表格图 · 12 个数据点从图片估算Chatbot Arena Elo ScoreGoogle · 柱状图 · 7 个数据点从图片估算Trending Food ItemsGoogle · 表格图 · 6 个数据点从图片估算Domestic FlightsGoogle · 表格图 · 2 个数据点从图片估算Gemini 3.8 Flash vs. Other ModelsGoogle · 表格图 · 102 个数据点从图片估算Al momentum redefining what's possible across the companyGoogle · 表格图 · 6 个数据点从图片估算Weather ConditionsGoogle · 表格图 · 3 个数据点
Meta 96
全部模型从图片估算相似重绘 8.3Llama3.1 405B Human EvaluationMeta · 堆叠柱状图 · 9 个数据点从图片估算相似重绘 8.1AI Model Performance TrendMeta · 折线图 · 4 个数据点从图片估算Terminal-Bench 2.1, DeepSWE 1.1, Meta Internal Coding BenchMeta · 跑分柱状图 · 17 个数据点从图片估算Google and Meta are Benchmaxxerssemianalysis · 表格图 · 12 个数据点从图片估算Benchmark ComparisonMeta · 表格图 · 13 个数据点从图片估算MCP Atlas vs GDPval-AA V2Meta · 跑分柱状图 · 14 个数据点从图片估算BenchmarkMeta · 表格图 · 54 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 柱状图 · 32 个数据点从图片估算相似重绘 7.1Artificial Analysis Intelligence Index v4.1artificial-analysis · 柱状图 · 37 个数据点从图片估算Benchmark ComparisonMeta · 表格图 · 78 个数据点从图片估算Artificial Analysis Intelligence Index v4.1.1artificial-analysis · 柱状图 · 54 个数据点从图片估算Sharp threshold for exact fit in high dimensionsMeta · 表格图 · 10 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 表格图 · 44 个数据点从图片估算相似重绘 7.8LMArena ELO score vs. costMeta · 散点图 · 13 个数据点从图片估算Held Out Codebase PerplexityMeta · 折线图 · 12 个数据点从图片估算相似重绘 8.4Improvement With Self-RefinementMeta · 堆叠柱状图 · 6 个数据点从图片估算Out-of-Distribution Performance vs In-Distribution PerformanceMeta · 散点图 · 8 个数据点从图片估算Improvement With Test-Time ComputeMeta · 折线图 · 10 个数据点从图片估算DiarizationMeta · 柱状图 · 6 个数据点从图片估算相似重绘 7.6MemoryBenchMeta · 柱状图 · 5 个数据点从图片估算FocusBenchMeta · 柱状图 · 5 个数据点从图片估算SI-SDR improvement vs Number of speakersMeta · 散点图 · 26 个数据点从图片估算Artificial Analysis Intelligence Index v4.3.2artificial-analysis · 分组柱状图 · 47 个数据点从图片估算Intelligence Index vs. Cost per Intelligence Index Taskartificial-analysis · 散点图 · 28 个数据点从图片估算Net Win Rate of SAM Audio vs. SoTA SeparationMeta · 分组柱状图 · 92 个数据点从图片估算Personal and professional agents benchmark comparisonMeta · 表格图 · 65 个数据点从图片估算Model Performance ComparisonMeta · 跑分柱状图 · 32 个数据点从图片估算DFlash Speculative Decoding PerformanceMeta · 跑分柱状图 · 6 个数据点从图片估算SAM 3.1 and SAM 3 with Realistic Workload on a Single November 2025 H100Meta · 折线图 · 16 个数据点从图片估算Code Muse Spark 1.3 (xHigh) Ranked ~#10arena · 柱状图 · 31 个数据点从图片估算AIRA3 - NVIDIA Nemotron Model Reasoning ChallengeMeta · 表格图 · 39 个数据点从图片估算Text-to-Image Arenaarena · 柱状图 · 15 个数据点从图片估算Artificial Analysis Intelligence Index v4.1.1artificial-analysis · 柱状图 · 40 个数据点从图片估算Circular network diagram of shift and generalisation attributesMeta · 表格图 · 17 个数据点从图片估算Cyber Misuse Refusal BenchmarkMeta · 柱状图 · 4 个数据点从图片估算Multimodal IntelligenceMeta · 折线图 · 4 个数据点从图片估算Mishandling Personal InformationMeta · 柱状图 · 4 个数据点从图片估算Text-to-Video Arena Ranks #3arena · 柱状图 · 12 个数据点从图片估算WideSearchMeta · 折线图 · 24 个数据点从图片估算AA-WER Streaming Index - Final TranscriptionMeta · 柱状图 · 9 个数据点从图片估算相似重绘 7.9Model Benchmark ComparisonMeta · 表格图 · 98 个数据点从图片估算Artificial Intelligence Index incorporates 7 evaluations: MMLU-Pro, GPQA Diamond, Humanity's Last Exam, LiveCodeBench, SciCode, AIME, MATH-500artificial-analysis · 柱状图 · 16 个数据点从图片估算Vibe Code Bench v1.1Meta · 跑分柱状图 · 4 个数据点从图片估算Exact vs Not exactMeta · 表格图 · 10 个数据点从图片估算Humanity's Last Exam (With Tools)Meta · 折线图 · 10 个数据点从图片估算Meta Internal Coding BenchMeta · 柱状图 · 5 个数据点从图片估算V-JEPA 2 vs V-JEPA 2.1 vs Previous SOTAMeta · 跑分柱状图 · 24 个数据点从图片估算Llama 3.1Meta · 柱状图 · 3 个数据点从图片估算Model Performance ComparisonMeta · 跑分柱状图 · 15 个数据点从图片估算Muse Image: Gap to Use Case Frontier (Text to Image)artificial-analysis · 表格图 · 30 个数据点从图片估算Muse Image: Gap to Editing Action Frontier (Image Editing)artificial-analysis · 表格图 · 21 个数据点从图片估算Artificial Analysis underlying Openness Index: Componentsartificial-analysis · 堆叠柱状图 · 80 个数据点从图片估算Cost per Intelligence Index Taskartificial-analysis · 跑分柱状图 · 35 个数据点从图片估算Artificial Analysis Coding Agent Index vs. average total tokens per taskartificial-analysis · 散点图 · 16 个数据点从图片估算Artificial Analysis Coding Agent Indexartificial-analysis · 跑分柱状图 · 31 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 柱状图 · 23 个数据点从图片估算CritE Scoreartificial-analysis · 分组柱状图 · 25 个数据点从图片估算Output Speed vs. Output Price: Llama 4 Maverick Providersartificial-analysis · 散点图 · 11 个数据点从图片估算Multi-Meta Image Edit Arena Ranks #2arena · 跑分柱状图 · 15 个数据点从图片估算Harvey LAB-AA v1.1: All-Pass Rateartificial-analysis · 分组柱状图 · 21 个数据点从图片估算Output Tokens per Second: Llama; Higher is Maverickartificial-analysis · 柱状图 · 9 个数据点从图片估算Output Speed vs. Price: Llama 4 Scoutartificial-analysis · 散点图 · 5 个数据点从图片估算Muse Image: Gap to Capability Frontier (Text to Image)artificial-analysis · 表格图 · 27 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 跑分柱状图 · 59 个数据点从图片估算Output Speed vs. Price: Llama 4 Scout Providersartificial-analysis · 散点图 · 10 个数据点从图片估算Context Window: Tokens Limit: Llama 4 Maverick Providersartificial-analysis · 柱状图 · 10 个数据点从图片估算Artificial Analysis Coding Agent Index vs. average pay-per-token API cost per task (USD)artificial-analysis · 散点图 · 26 个数据点从图片估算Output Tokens Used to Run Artificial Analysis Intelligence Indexartificial-analysis · 堆叠柱状图 · 50 个数据点从图片估算Pricing per 1M Tokens: Llama 4 Scout Providersartificial-analysis · 分组柱状图 · 20 个数据点从图片估算Image Meta Muse Image Ranks #2arena · 柱状图 · 15 个数据点从图片估算Cost per Taskartificial-analysis · 柱状图 · 15 个数据点从图片估算Presentation Elo vs. Rubric Score (%)artificial-analysis · 散点图 · 26 个数据点从图片估算Output Tokens per Second: Llama 4 Maverick Providersartificial-analysis · 柱状图 · 9 个数据点从图片估算GDPval-AAv2 Leaderboardartificial-analysis · 跑分柱状图 · 37 个数据点从图片估算Training compute (FLOP) of notable open modelsepoch-ai · 散点图 · 27 个数据点从图片估算Output Speed vs. Price: Llama 4 Maverick Providersartificial-analysis · 散点图 · 4 个数据点从图片估算Agent Arena Muse Spark 1.3 (Max)arena · 表格图 · 6 个数据点从图片估算Agent Arena Spark 1.3 (Max): Ranked #13arena · 柱状图 · 25 个数据点从图片估算τ³-Benchmarking: Scoreartificial-analysis · 柱状图 · 39 个数据点从图片估算USD per 1M Tokens: Input and Output Prices: Llama 4 Maverick Providersartificial-analysis · 分组柱状图 · 18 个数据点从图片估算Agent Arena Muse Spark 1.2 (xHigh)arena · 表格图 · 6 个数据点从图片估算Agent Arena Ranked #25arena · 柱状图 · 28 个数据点从图片估算Text Arena - Open Weight Rank #24arena · 柱状图 · 31 个数据点从图片估算Muse Image: Gap to Use Case Frontier (Image Editing)artificial-analysis · 表格图 · 30 个数据点从图片估算Intelligence Evaluationsartificial-analysis · 跑分柱状图 · 275 个数据点从图片估算Code Muse Arena - Open Weight Rank #26arena · 跑分柱状图 · 30 个数据点从图片估算Pareto Frontier for Code Arenaarena · 散点图 · 29 个数据点从图片估算GPQA Accuracies for Leading Modelsepoch-ai · 分组柱状图 · 8 个数据点从图片估算Context Window: Tokens Limit; Llama 4 Higher is better Providersartificial-analysis · 柱状图 · 10 个数据点从图片估算META AI MODEL RESULTSMeta · 表格图 · 5 个数据点从图片估算Open Materials 2024 (OMat24)Meta · 折线图 · 18 个数据点从图片估算Built with LlamaMeta · 表格图 · 10 个数据点从图片估算AIRA3 - NVIDIA Nemotron Model Reasoning ChallengeMeta · 折线图 · 97 个数据点从图片估算BenchmarkMeta · 表格图 · 47 个数据点从图片估算Live Evaluation ResultsMeta · 跑分柱状图 · 26 个数据点从图片估算Wild Artifact BenchMeta · 散点图 · 21 个数据点
xAI 108
全部模型从图片估算相似重绘 8.1Humanity's Last ExamxAI · 柱状图 · 7 个数据点从图片估算CaseLaw BenchmarkxAI · 表格图 · 6 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 17 个数据点从图片估算CursorBench 4.0 scorexAI · 折线图 · 17 个数据点从图片估算Grok-4.20-Beta1 ranks #1 in Search Arenaarena · 表格图 · 40 个数据点从图片估算相似重绘 5.3Artificial Analysis Intelligence Index v4.1.1artificial-analysis · 柱状图 · 29 个数据点从图片估算相似重绘 8.1Artificial Analysis Intelligence Index v4.1artificial-analysis · 分组柱状图 · 30 个数据点从图片估算AI Benchmark ComparisonxAI · 跑分柱状图 · 46 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 40 个数据点从图片估算AI Model Benchmarksartificial-analysis · 跑分柱状图 · 120 个数据点从图片估算xAI Grok-3 #1 in Arenaarena · 表格图 · 13 个数据点从图片估算相似重绘 8.1Intelligence vs. PricexAI · 散点图 · 17 个数据点从图片估算Scaling HLE - Test timexAI · 散点图 · 28 个数据点从图片估算AA-Briefcase Eloartificial-analysis · 柱状图 · 35 个数据点从图片估算Intelligence vs. PricexAI · 散点图 · 9 个数据点从图片估算Code Arena: Ranked #3arena · 柱状图 · 27 个数据点从图片估算r²-bench TelecomxAI · 散点图 · 2 个数据点从图片估算AA-WER Streaming Index - Final Transcription after detected End of Speechartificial-analysis · 分组柱状图 · 41 个数据点从图片估算Hallucination Rate vs FActScorexAI · 跑分柱状图 · 4 个数据点从图片估算Gemini 2.5 Flash Leaderboardarena · 表格图 · 45 个数据点从图片估算Artificial Analysis Coding Agent Index vs. Cost per Taskartificial-analysis · 跑分柱状图 · 22 个数据点从图片估算Grok 4.1 vs. previous GrokxAI · 表格图 · 1 个数据点从图片估算Big Bench Audio: Intelligence vs LatencyxAI · 散点图 · 1 个数据点从图片估算Model PerformancexAI · 散点图 · 1 个数据点从图片估算Artificial Analysis Speech to Speech Indexartificial-analysis · 跑分柱状图 · 35 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 柱状图 · 19 个数据点从图片估算Cost per minutexAI · 柱状图 · 5 个数据点从图片估算Grok-3 is also #1 across all categoriesarena · 表格图 · 120 个数据点从图片估算Artificial Analysis Intelligence Index vs. Priceartificial-analysis · 散点图 · 15 个数据点从图片估算Berkeley Function Calling v4 BenchmarkxAI · 散点图 · 6 个数据点从图片估算Speech Think-to-Speech IndexxAI · 柱状图 · 8 个数据点从图片估算Artificial Analysis Cyber Indexartificial-analysis · 跑分柱状图 · 48 个数据点从图片估算Artificial Analysis Intelligence Index v3.0artificial-analysis · 柱状图 · 35 个数据点从图片估算EQ-Emotional Intelligence Benchmark - Elo (Normalized)xAI · 柱状图 · 8 个数据点从图片估算Agent Arenaarena · 柱状图 · 19 个数据点从图片估算Speech Transcription: Accuracy vs. Price (Streaming)xAI · 散点图 · 11 个数据点从图片估算Instruction followingxAI · 柱状图 · 8 个数据点从图片估算xAI Grok-4 tied for #3 in Text Arenaarena · 表格图 · 33 个数据点从图片估算Text-to-Image Arena Top 20arena · 柱状图 · 20 个数据点从图片估算Grok-4-fast-search lands at #1arena · 表格图 · 43 个数据点从图片估算相似重绘 8.1AI Voice Model ComparisonxAI · 表格图 · 19 个数据点从图片估算LMArena Text LeaderboardxAI · 柱状图 · 16 个数据点从图片估算Claude Opus-4-6 & Sonnet 4.6 ranks #1 & #7 in Search Arenaarena · 表格图 · 19 个数据点从图片估算Creative Writing v3xAI · 柱状图 · 7 个数据点从图片估算Artificial Analysis Intelligence Index (Version 2, released Feb '25); Higher is betterartificial-analysis · 柱状图 · 21 个数据点从图片估算Model Performance ComparisonxAI · 跑分柱状图 · 6 个数据点从图片估算Telephony (8 kHz)xAI · 分组柱状图 · 32 个数据点从图片估算Gemini 2.5 Pro #1 on Arena with the largest score jump ever (+40)!arena · 表格图 · 17 个数据点从图片估算Artificial Analysis Text to Video Leaderboardartificial-analysis · 表格图 · 30 个数据点从图片估算Output Tokens used Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 63 个数据点从图片估算Model Rankingsarena · 表格图 · 7 个数据点从图片估算Intelligence Analysis vs. Cost to Run Artificial Analysis Intelligence Indexartificial-analysis · 散点图 · 20 个数据点从图片估算Average time to First audio (seconds) on Big Bench Audio dataset - Lower is betterartificial-analysis · 柱状图 · 29 个数据点从图片估算Agent Arenaarena · 柱状图 · 6 个数据点从图片估算Artificial Analysis Word Error Rate Index (Non-streaming) vs Priceartificial-analysis · 散点图 · 32 个数据点从图片估算Price per 1000 minutes of audioartificial-analysis · 柱状图 · 38 个数据点从图片估算Code Arena: WebDev Ranked #10arena · 柱状图 · 24 个数据点从图片估算Intelligence Analysis Index vs. Cost per Task (USD)artificial-analysis · 散点图 · 28 个数据点从图片估算Time per Intelligence Index Taskartificial-analysis · 柱状图 · 26 个数据点从图片估算Artificial Reasoning, Conversational Speech to Speech and Agentic Index Performance by Datasetartificial-analysis · 跑分柱状图 · 42 个数据点从图片估算Output Tokens per Intelligence Indexartificial-analysis · 跑分柱状图 · 81 个数据点从图片估算Output Tokens Speed per Second; Higher is betterartificial-analysis · 柱状图 · 21 个数据点从图片估算AA-WER Streaming Index vs. Time to Final Transcriptionartificial-analysis · 散点图 · 24 个数据点从图片估算AA-WER: % of words transcribed incorrectly on the Earnings22-Cleaned-AA-v1 Datasetartificial-analysis · 柱状图 · 36 个数据点从图片估算AA-Briefcase Elo vs. Cost per Taskartificial-analysis · 散点图 · 42 个数据点从图片估算FrontierMath accuracy for Grok-3 and Grok-3 mini compared to leading modelsepoch-ai · 柱状图 · 9 个数据点从图片估算Artificial Analysis Word Error Rate (AA-WER) Indexartificial-analysis · 柱状图 · 33 个数据点从图片估算Grok Imagine Image 2.0: Gap to Capability Frontier (Text to Image)artificial-analysis · 表格图 · 30 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 柱状图 · 39 个数据点从图片估算Total Training FLOPs by Modelepoch-ai · 折线图 · 6 个数据点从图片估算Output Tokens Used to Run Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 46 个数据点从图片估算Grok Imagine Image 2.0: Gap to Editing Action Frontier (Image Editing)artificial-analysis · 表格图 · 21 个数据点从图片估算Grok Imagine Video 1.5 Lite: Gap to Capability Frontier (AA-Video-T2V v2.0)artificial-analysis · 表格图 · 30 个数据点从图片估算AA-Briefcase Elo vs. Time per Taskartificial-analysis · 散点图 · 47 个数据点从图片估算GDPval-AA v2 Leaderboardartificial-analysis · 柱状图 · 32 个数据点从图片估算Training compute of notable modelsepoch-ai · 散点图 · 7 个数据点从图片估算Agent Arena: Grok over timearena · 折线图 · 8 个数据点从图片估算End-to-End Output 500 Response Timeartificial-analysis · 分组柱状图 · 58 个数据点从图片估算Image-to-Video Arenaarena · 柱状图 · 16 个数据点从图片估算Grok Imagine Video 1.5 Lite: Quality vs Speed Frontier (AA-Video-T2V-Silent v2.0)artificial-analysis · 散点图 · 9 个数据点从图片估算Image Edit Arena Top 20arena · 柱状图 · 19 个数据点从图片估算AA-WER Streaming Index (First Partial) vs. Time to First Partial Transcription After Speech Endartificial-analysis · 散点图 · 21 个数据点从图片估算相似重绘 5.3Artificial Analysis Intelligence Index V4.3artificial-analysis · 柱状图 · 29 个数据点从图片估算Artificial Analysis Coding Agent Indexartificial-analysis · 跑分柱状图 · 69 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 柱状图 · 36 个数据点从图片估算Grok Imagine Video 1.5 Lite: Gap to Use Case Frontierartificial-analysis · 表格图 · 30 个数据点从图片估算Grok Imagine Image 2.0: Gap to Use Case Frontier (Text to Image)artificial-analysis · 表格图 · 20 个数据点从图片估算Text-to-Video Arenaarena · 柱状图 · 15 个数据点从图片估算AA-Briefcase Eloartificial-analysis · 分组柱状图 · 27 个数据点从图片估算Artificial Analysis Coding Agent Index (Composite average pass@1 across DeepSWE, Terminal-Bench v2, and SWE-Atlas-QnA)artificial-analysis · 分组柱状图 · 20 个数据点从图片估算Artificial Analysis Intelligence Index v3.0artificial-analysis · 跑分柱状图 · 32 个数据点从图片估算Code Arena: WebDevarena · 柱状图 · 15 个数据点从图片估算AA-WER Streaming Index (First Partial) vs. Time to First Partial Transcription After Speech Endartificial-analysis · 散点图 · 32 个数据点从图片估算AA-Briefcase Eloartificial-analysis · 分组柱状图 · 42 个数据点从图片估算Intelligence Evaluationsartificial-analysis · 柱状图 · 100 个数据点从图片估算DeepSWE v1.1 Benchmark Scoreartificial-analysis · 分组柱状图 · 19 个数据点从图片估算Grok Imagine Image 2.0: Gap to Use Case Frontier (Image Editing)artificial-analysis · 表格图 · 20 个数据点从图片估算Language Model Inference Priceartificial-analysis · 折线图 · 80 个数据点从图片估算Cost per complete fixed 40-question Big Audio Bench (Big Bench Audio subset based on length of input audio, normalized to hourly basis - Lower is better)artificial-analysis · 柱状图 · 21 个数据点从图片估算SWE-Atlas-QnA Benchmark ScorexAI · 柱状图 · 14 个数据点从图片估算Token usage by modelxAI · 表格图 · 110 个数据点从图片估算Token usage LeaderboardxAI · 堆叠柱状图 · 66 个数据点从图片估算Top chartsxAI · 表格图 · 6 个数据点从图片估算TSLA Tesla, Inc.xAI · 折线图 · 5 个数据点从图片估算CategoriesxAI · 堆叠柱状图 · 9 个数据点从图片估算VEHICLE SUMMARYxAI · 表格图 · 44 个数据点从图片估算FrontierCode 1.1 Main Score vs CostxAI · 散点图 · 16 个数据点从图片估算Voting intentionxAI · 柱状图 · 10 个数据点
DeepSeek 97
全部模型部分为发布方的数字相似重绘 8.1Accuracy / Percentile (%)DeepSeek · 跑分柱状图 · 30 个数据点从图片估算Global KV Cache Per Token (Bytes)DeepSeek · 柱状图 · 4 个数据点从图片估算相似重绘 6.3AI Model Benchmark Performance Comparisonqwen · 跑分柱状图 · 25 个数据点部分为发布方的数字相似重绘 8.3Accuracy / Percentile (%)DeepSeek · 跑分柱状图 · 30 个数据点部分为发布方的数字Accuracy / Percentile (%)DeepSeek · 跑分柱状图 · 36 个数据点从图片估算相似重绘 9.0AI Model Benchmarks ComparisonDeepSeek · 跑分柱状图 · 20 个数据点从图片估算相似重绘 7.4AI Model Benchmarks Comparisonqwen · 跑分柱状图 · 25 个数据点从图片估算相似重绘 9.0Model Performance Comparison Across Benchmarksgithub:zai-org/GLM-4 · 跑分柱状图 · 30 个数据点部分为发布方的数字相似重绘 8.2Accuracy Percentile by Model and BenchmarkDeepSeek · 跑分柱状图 · 25 个数据点从图片估算相似重绘 8.6Artificial Analysis Intelligence Index incorporates 7 evaluations: MMLU-Pro, GPQA Diamond, Humanity's Last Exam, LiveCodeBench, SciCode, AIME, MATH-500artificial-analysis · 柱状图 · 16 个数据点从图片估算Artificial Analysis Intelligence Index (Non-Reasoning Models)artificial-analysis · 柱状图 · 17 个数据点部分为发布方的数字Accuracy / Pass@1 (%)DeepSeek · 跑分柱状图 · 28 个数据点从图片估算相似重绘 7.3Model Performance Comparison Across Benchmarksgithub:zai-org/GLM-4 · 跑分柱状图 · 50 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 63 个数据点从图片估算Benchmark Comparison of Reasoning and Agentic CapabilitiesDeepSeek · 跑分柱状图 · 44 个数据点从图片估算Cost Per Million Tokens vs Token PositionDeepSeek · 折线图 · 20 个数据点从图片估算Pressure Testing DeepSeek-V3 128K Context via "Needle In A HayStack"DeepSeek · 表格图 · 15 个数据点从图片估算DeepSeek Model Performance ComparisonDeepSeek · 折线图 · 30 个数据点从图片估算Pressure Testing DeepSeek-Coder-V2 Base 128K Context via "Needle In A HayStack"DeepSeek · 表格图 · 11 个数据点从图片估算deepseek-r1-lite-preview AIME accuracy with test-time compute scalingDeepSeek · 折线图 · 17 个数据点从图片估算Pass Rate ComparisonDeepSeek · 跑分柱状图 · 14 个数据点从图片估算Agent DeepSeek-V4.1-Flash (Max): Ranked #3 Openarena · 分组柱状图 · 15 个数据点从图片估算Average PerformanceDeepSeek · 表格图 · 9 个数据点从图片估算Inference costs of DeepSeek-V3.1-Terminus and DeepSeek-V3.2-ExpDeepSeek · 折线图 · 20 个数据点从图片估算(a) Compression on Fox benchmarkDeepSeek · 表格图 · 28 个数据点从图片估算Code DeepSeek Arena Top3 Openarena · 柱状图 · 15 个数据点从图片估算Pass Rate ComparisonDeepSeek · 跑分柱状图 · 30 个数据点从图片估算Model Performance on Three BenchmarksDeepSeek · 跑分柱状图 · 15 个数据点部分为发布方的数字AIME 2024DeepSeek · 柱状图 · 6 个数据点从图片估算Agent DeepSeekArena-V4-Flash (High)arena · 柱状图 · 30 个数据点从图片估算Programming Language Performance ComparisonDeepSeek · 表格图 · 48 个数据点从图片估算Benchmark PerformancesDeepSeek · 表格图 · 45 个数据点从图片估算LeaderboardDeepSeek · 散点图 · 23 个数据点从图片估算MMLU Redux ZeroEval Score VS Input API Price ($ / 1M Tokens)DeepSeek · 散点图 · 13 个数据点从图片估算Artificial Analysis Intelligence Index v2artificial-analysis · 柱状图 · 22 个数据点从图片估算DeepSeek Model BenchmarksDeepSeek · 跑分柱状图 · 12 个数据点从图片估算Human evaluationsDeepSeek · 跑分柱状图 · 20 个数据点部分为发布方的数字Accuracy / Pass@1 (%)DeepSeek · 跑分柱状图 · 31 个数据点从图片估算Benchmark PerformanceDeepSeek · 表格图 · 36 个数据点从图片估算AI Model Benchmark Performance ComparisonDeepSeek · 跑分柱状图 · 20 个数据点从图片估算Frontend DeepSeek-V4 Flash Arena (High)arena · 柱状图 · 19 个数据点从图片估算DeepSeek V4 Flash 0731 debuts as the second-strongest open-weights modelepoch-ai · 跑分柱状图 · 11 个数据点从图片估算Pareto Frontier for Frontend Code Arenaarena · 折线图 · 6 个数据点从图片估算Interpolated Token Throughput per GPU (tok/s/gpu)semianalysis · 折线图 · 85 个数据点从图片估算Intelligence Parameters vs. Active Parametersartificial-analysis · 散点图 · 15 个数据点从图片估算AI Model Accuracy Comparison Across BenchmarksDeepSeek · 跑分柱状图 · 42 个数据点从图片估算MATH Top@1 Accuracy over TimeDeepSeek · 折线图 · 7 个数据点部分为发布方的数字相似重绘 7.9Accuracy / Percentile (%) by BenchmarkDeepSeek · 跑分柱状图 · 30 个数据点从图片估算Average Performance vs Activated ParametersDeepSeek · 散点图 · 10 个数据点部分为发布方的数字AIME 2024DeepSeek · 柱状图 · 8 个数据点从图片估算相似重绘 7.8Input/Output Pricing for o1-Class Inference Models ($/1M Tokens)DeepSeek · 分组柱状图 · 12 个数据点从图片估算Artificial Analysis Intelligence Index v4.3artificial-analysis · 表格图 · 24 个数据点从图片估算Agent DeepSeek-V4-Pro (High)arena · 柱状图 · 16 个数据点从图片估算Artificial Analysis Intelligence Index v4.1 by Open Weights / Proprietaryartificial-analysis · 跑分柱状图 · 56 个数据点从图片估算Win Rates Across Website Designarena · 分组柱状图 · 32 个数据点从图片估算GDPval-AA Models Leaderboardartificial-analysis · 柱状图 · 16 个数据点从图片估算AA-Briefcase rubric Score from binary rubric checks (%)artificial-analysis · 柱状图 · 30 个数据点从图片估算GDPval-AA Leaderboardartificial-analysis · 跑分柱状图 · 52 个数据点从图片估算GDPval-AA v2artificial-analysis · 分组柱状图 · 35 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 柱状图 · 28 个数据点从图片估算Text DeepSeek:V4-Top15 (Max)arena · 柱状图 · 16 个数据点从图片估算Output Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 56 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 分组柱状图 · 25 个数据点从图片估算Jev Routerarena · 柱状图 · 11 个数据点从图片估算Code DeepSeek Arena: V4-Pro (Max) Top15arena · 柱状图 · 15 个数据点从图片估算Output Tokens per Second: DeepSeek V3.2 Exp; Higher is better; 1,000 Input Providersartificial-analysis · 柱状图 · 4 个数据点从图片估算Artificial Analysis Coding Indexartificial-analysis · 柱状图 · 32 个数据点从图片估算Elo rating-AA for performance on real-world work tasksartificial-analysis · 柱状图 · 28 个数据点从图片估算Top 7 Open Models: February 2026 in Code Arenaarena · 柱状图 · 7 个数据点部分为发布方的数字Intelligence Analysis vs. Intelligence Index; Output Tokens Used in Artificial Analysis Intelligence Indexartificial-analysis · 散点图 · 8 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 柱状图 · 27 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 分组柱状图 · 24 个数据点从图片估算GDPval-AA v2 Leaderboardartificial-analysis · 柱状图 · 35 个数据点从图片估算Intelligence Evaluationsartificial-analysis · 柱状图 · 100 个数据点从图片估算AutomationBench-AA: Score vs. Cost per Taskartificial-analysis · 散点图 · 22 个数据点从图片估算Output Tokensartificial-analysis · 跑分柱状图 · 58 个数据点从图片估算Text Arena Score by Categoryarena · 表格图 · 70 个数据点从图片估算Intelligence vs. Cost to Run Artificial Intelligence Indexartificial-analysis · 散点图 · 32 个数据点从图片估算Agent DeepSeek-V4-Pro (High)arena · 跑分柱状图 · 12 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 表格图 · 34 个数据点从图片估算Pareto Frontier for Agent Arenaarena · 散点图 · 9 个数据点从图片估算Output Reasoning Tokens Used to Run Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 23 个数据点从图片估算Output Tokens per Second: DeepSeek V3.1artificial-analysis · 柱状图 · 5 个数据点从图片估算Agent DeepSeek Arena Pareto Frontierarena · 散点图 · 9 个数据点从图片估算Agent DeepSeek-Arena-V4-Flash: Rank #21arena · 柱状图 · 6 个数据点从图片估算Win Rates Across Game Developmentarena · 跑分柱状图 · 28 个数据点从图片估算Output Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 50 个数据点从图片估算Cost per Intelligence Indexartificial-analysis · 跑分柱状图 · 130 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 跑分柱状图 · 59 个数据点从图片估算AA-Briefcase Eloartificial-analysis · 柱状图 · 39 个数据点从图片估算DeepSeek-V4.1-Flash (Max) Pareto Frontier for Agent Arenaarena · 折线图 · 14 个数据点从图片估算Intelligence Evaluationsartificial-analysis · 跑分柱状图 · 107 个数据点从图片估算Intelligence vs. Total Parametersartificial-analysis · 散点图 · 19 个数据点从图片估算DeepSeek V4 Pro (Max) Pareto Frontier for Code Arenaarena · 散点图 · 8 个数据点从图片估算Artificial Analysis Intelligence Index by Open Weights / Proprietaryartificial-analysis · 柱状图 · 24 个数据点从图片估算DeepSeek-V4 API New PricingDeepSeek · 表格图 · 12 个数据点部分为发布方的数字Benchmark (Metric) ComparisonDeepSeek · 表格图 · 111 个数据点
Qwen 117
全部模型从图片估算AI Model Benchmark Performance ComparisonQwen · 跑分柱状图 · 25 个数据点从图片估算相似重绘 7.8AI Model Benchmarks ComparisonQwen · 跑分柱状图 · 25 个数据点从图片估算Model Comparison Across BenchmarksQwen · 跑分柱状图 · 92 个数据点从图片估算Qwen3.8-Max: Ranks #2arena · 分组柱状图 · 19 个数据点从图片估算Model BenchmarksQwen · 跑分柱状图 · 55 个数据点从图片估算Model Benchmark ComparisonQwen · 表格图 · 58 个数据点从图片估算Model Benchmark ComparisonQwen · 表格图 · 63 个数据点从图片估算Model Benchmarks ComparisonQwen · 跑分柱状图 · 85 个数据点从图片估算相似重绘 7.2Model Comparison BenchmarksQwen · 跑分柱状图 · 84 个数据点从图片估算Pareto FrontierQwen · 散点图 · 7 个数据点从图片估算Model Comparison Across BenchmarksQwen · 跑分柱状图 · 120 个数据点从图片估算Code Arena 3.7 Max (20260517) : #3arena · 柱状图 · 19 个数据点从图片估算相似重绘 7.4Frontend3 Code Max: Ranks #4arena · 柱状图 · 19 个数据点从图片估算相似重绘 7.7Code Arena: WebDevarena · 柱状图 · 14 个数据点从图片估算Qwen3.8-Max Pareto Frontier for Frontend Code Arenaarena · 折线图 · 7 个数据点从图片估算Relative Prefill Throughput at 90% Cache Hit RateQwen · 折线图 · 21 个数据点从图片估算ASR Error Rates - Public TestQwen · 分组柱状图 · 29 个数据点从图片估算相似重绘 8.4Qwen3-Coder-Plus Benchmark ComparisonQwen · 跑分柱状图 · 8 个数据点从图片估算Artificial Analysis Intelligence IndexQwen · 柱状图 · 43 个数据点从图片估算Average Ranking vs. Environment ScalingQwen · 折线图 · 14 个数据点从图片估算Qwen3-Coder-Flash BenchmarksQwen · 表格图 · 135 个数据点从图片估算Qwen3-TTS-12Hz-1.7B ComparisonQwen · 表格图 · 58 个数据点从图片估算相似重绘 7.6Model Performance ComparisonQwen · 跑分柱状图 · 25 个数据点从图片估算Artificial Analysis Intelligence Agentic IndexQwen · 柱状图 · 31 个数据点从图片估算SWE-bench Performance Verified (%)Qwen · 散点图 · 11 个数据点从图片估算VramQwen · 柱状图 · 2 个数据点从图片估算Model Benchmarks ComparisonQwen · 跑分柱状图 · 79 个数据点从图片估算Model Performance Comparison Across BenchmarksQwen · 跑分柱状图 · 42 个数据点从图片估算Qwen3-Coder BenchmarksQwen · 表格图 · 135 个数据点从图片估算Model Performance Comparison Across BenchmarksQwen · 跑分柱状图 · 78 个数据点从图片估算Qwen 3.6 Max (preview) Benchmark ComparisonQwen · 跑分柱状图 · 70 个数据点从图片估算Total images processed on OpenRouterQwen · 堆叠柱状图 · 70 个数据点从图片估算Qwen2.5-Omni Performance ComparisonQwen · 跑分柱状图 · 10 个数据点从图片估算Model Benchmarks ComparisonQwen · 跑分柱状图 · 66 个数据点从图片估算AI Model Benchmark ComparisonQwen · 跑分柱状图 · 25 个数据点从图片估算Qwen3-Omni-Flash2025-12-01 vs Qwen3-Omni-Flash vs Qwen3-235B-A22B vs GPT-4o vs Gemini-2.5-FlashQwen · 表格图 · 158 个数据点从图片估算AI Model Benchmark Performance ComparisonQwen · 跑分柱状图 · 25 个数据点从图片估算Context LengthQwen · 跑分柱状图 · 6 个数据点从图片估算Qwen3.5-Omni-Plus Benchmark ComparisonQwen · 表格图 · 76 个数据点从图片估算Content Consistency (↓) - TTS Multilingual Test SetQwen · 跑分柱状图 · 40 个数据点从图片估算Model Benchmarks ComparisonQwen · 表格图 · 223 个数据点从图片估算Content Consistency (↓) - TTS Multilingual Test SetQwen · 分组柱状图 · 65 个数据点从图片估算Model Benchmarks ComparisonQwen · 跑分柱状图 · 69 个数据点从图片估算Performance on Coding Agent BenchmarksQwen · 跑分柱状图 · 20 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 柱状图 · 37 个数据点从图片估算Qwen-Image-2.1: Gap to Open Weights Use Case Frontierartificial-analysis · 表格图 · 20 个数据点从图片估算Qwen-RobotNav Benchmark ResultsQwen · 跑分柱状图 · 57 个数据点从图片估算Chatbot Arena Category: Overall vs Creative Writingarena · 折线图 · 12 个数据点从图片估算Total Tokens per $1 TCO (Owning - Hyperscaler) vs. P90 Interactivitysemianalysis · 散点图 · 66 个数据点从图片估算Qwen-Image-2.1: Gap to Editing Action Frontier (AA-Image-Editing v2.0)artificial-analysis · 表格图 · 21 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 表格图 · 30 个数据点从图片估算Output Tokens per Second: Qwen2.5 72Bartificial-analysis · 柱状图 · 7 个数据点从图片估算Wan 2.5-i2i-Preview ranks #21 in Image Editarena · 表格图 · 18 个数据点从图片估算Qwen3-ASR-1.7B vs Other ASR ModelsQwen · 表格图 · 90 个数据点从图片估算Intelligence Analysis vs. Cost to Run Intelligence Indexartificial-analysis · 散点图 · 31 个数据点从图片估算Qwen2.5-Omni vs Other Models Across BenchmarksQwen · 跑分柱状图 · 21 个数据点从图片估算Text-to-Image Model Elo LeaderboardQwen · 表格图 · 55 个数据点从图片估算Qwen-Image-2.1: Gap to Use Case Frontier (AA-Image-T2I v2.0)artificial-analysis · 表格图 · 30 个数据点从图片估算Time to First Audioartificial-analysis · 柱状图 · 13 个数据点从图片估算Intelligence vs. Active Parameters: Qwen3 Modelsartificial-analysis · 散点图 · 16 个数据点从图片估算Model Performance ComparisonQwen · 跑分柱状图 · 48 个数据点从图片估算Wan #3 i2v preview ranks #3 in Image-to-Videoarena · 表格图 · 48 个数据点从图片估算Pareto Frontier for Code Arena: WebDevarena · 折线图 · 11 个数据点从图片估算Output Tokens per Second, Code Oriented Prompt: Qwen2.5 Coder 32Bartificial-analysis · 柱状图 · 5 个数据点从图片估算Qwen-Image-2.1: Gap to Open Weights Editing Action Frontierartificial-analysis · 表格图 · 21 个数据点从图片估算Qwen-Image-2.1: Gap to Open Weights Capability Frontier (AA-Image-T2I v2.0)artificial-analysis · 表格图 · 27 个数据点从图片估算Qwen3-VL Benchmark ComparisonQwen · 表格图 · 70 个数据点从图片估算Content Consistency (↓) - TTS Multilingual Test SetQwen · 分组柱状图 · 44 个数据点从图片估算Text-to-Image-2.1 Arena: Ranked #1 Openarena · 柱状图 · 10 个数据点从图片估算Completion Time by Model and Systemartificial-analysis · 表格图 · 14 个数据点从图片估算Image-to-WebDev Arena: Open Qwen-3.8-27B: Ranked #1arena · 柱状图 · 10 个数据点从图片估算Model ComparisonQwen · 表格图 · 209 个数据点从图片估算Output Tokens per Second: Qwen3 235B 2507 (Reasoning) Providersartificial-analysis · 柱状图 · 6 个数据点从图片估算(a) Task & Scene GeneralizationQwen · 分组柱状图 · 8 个数据点从图片估算Artificial Analysis Intelligence Index v4.0, <40B Total Parametersartificial-analysis · 柱状图 · 23 个数据点从图片估算Artificial Analysis Intelligence Index v3.0artificial-analysis · 柱状图 · 25 个数据点从图片估算Qwen-Image-Edit model. Ranks the #6 open overall.arena · 表格图 · 9 个数据点从图片估算Pareto Frontier for Code Arena: WebDevarena · 折线图 · 10 个数据点从图片估算Qwen-Image-2.1: Gap to Open Weights Use Case Frontier (AA-Image-Editing v2.0)artificial-analysis · 表格图 · 30 个数据点从图片估算Output Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 42 个数据点从图片估算Average Score (16K max context) vs. Output Tokensartificial-analysis · 散点图 · 21 个数据点从图片估算Qwen3-Max-Instruct scores much higher on AA-SimpleQA Omniscience Verified would predict its score onepoch-ai · 散点图 · 13 个数据点从图片估算Text-to-Image Arenaarena · 柱状图 · 14 个数据点从图片估算Cross-Harness GeneralizationQwen · 跑分柱状图 · 12 个数据点从图片估算Instruction Following (↑) - InstructTTS EvalQwen · 跑分柱状图 · 24 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 柱状图 · 28 个数据点从图片估算Image Edit Model Rankingsarena · 表格图 · 15 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 43 个数据点从图片估算Intelligence Analysis vs. Total Intelligence Indexartificial-analysis · 散点图 · 26 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 柱状图 · 41 个数据点从图片估算Intelligence vs. Active Parameters; Artificial Analysis Intelligence Indexartificial-analysis · 散点图 · 29 个数据点从图片估算Qwen-Image-2.1: Gap to Capability Frontier (AA-Image-T2I v2.0)artificial-analysis · 表格图 · 18 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 分组柱状图 · 22 个数据点从图片估算Qwen3-ASR-0.6B vs Other ASR ModelsQwen · 表格图 · 19 个数据点从图片估算Model ComparisonQwen · 表格图 · 135 个数据点从图片估算Total Tokens per $1 TCO (Owning - Hyperscaler Inference) vs. P90 Time To First Tokensemianalysis · 散点图 · 47 个数据点从图片估算Output Tokens per Second: Qwen3 Coder 480B Providersartificial-analysis · 柱状图 · 10 个数据点从图片估算Qwen3-VL vs Other ModelsQwen · 表格图 · 148 个数据点从图片估算Qwen Image 2.0 Pro ranks #9 for Image Arenaarena · 表格图 · 10 个数据点从图片估算Text-to-Image 3.0-Pro vs. 2.0-Pro Arena Score by Categoryarena · 表格图 · 2 个数据点从图片估算Intelligence vs. Output Tokens Used in Artificial Analysis Intelligence Indexartificial-analysis · 散点图 · 26 个数据点从图片估算Image Edit Arenaarena · 柱状图 · 10 个数据点从图片估算Qwen3-VL 235B-A22B Instruct vs Other ModelsQwen · 柱状图 · 42 个数据点从图片估算Qwen-Image-2.1: Gap to Use Case Frontier (AA-Image-Editing v2.0)artificial-analysis · 表格图 · 20 个数据点从图片估算Qwen Image 2512 is #2 open and #13 overall in Text-to-Imagearena · 表格图 · 36 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 柱状图 · 27 个数据点从图片估算ASR Domain Term Recognition AccuracyQwen · 表格图 · 27 个数据点从图片估算GPU utilization (%)Qwen · 表格图 · 50 个数据点从图片估算MobileEval BenchmarksQwen · 表格图 · 5 个数据点从图片估算Model Performance ComparisonQwen · 跑分柱状图 · 22 个数据点从图片估算(1) Building the Foundation for Agentic Environment SimulationQwen · 分组柱状图 · 19 个数据点从图片估算Qwen3-Omni ArchitectureQwen · 表格图 · 16 个数据点从图片估算Best-of-8 Accuracy vs ProcessBench Mean F1Qwen · 跑分柱状图 · 24 个数据点从图片估算Qwen2.5-Coder-32BQwen · 表格图 · 55 个数据点从图片估算AI model leaderboardQwen · 表格图 · 70 个数据点从图片估算Agent: Faster, More Cost-EffectiveQwen · 柱状图 · 2 个数据点从图片估算Text ArenaQwen · 表格图 · 58 个数据点
Kimi (Moonshot AI) 85
全部模型从图片估算相似重绘 9.0Model Performance ComparisonKimi (Moonshot AI) · 跑分柱状图 · 15 个数据点从图片估算Model BenchmarksKimi (Moonshot AI) · 跑分柱状图 · 47 个数据点从图片估算相似重绘 7.9Coding and Agents BenchmarksKimi (Moonshot AI) · 跑分柱状图 · 24 个数据点从图片估算Model Performance ComparisonKimi (Moonshot AI) · 跑分柱状图 · 38 个数据点从图片估算Internal Knowledge Work BenchKimi (Moonshot AI) · 跑分柱状图 · 9 个数据点从图片估算相似重绘 6.3Frontend Code Kimi-K3: Ranked #1arena · 柱状图 · 23 个数据点从图片估算OmniDocBench Scores by CategoryKimi (Moonshot AI) · 跑分柱状图 · 44 个数据点从图片估算相似重绘 3.5General AgentsKimi (Moonshot AI) · 跑分柱状图 · 45 个数据点从图片估算CodingKimi (Moonshot AI) · 跑分柱状图 · 36 个数据点从图片估算Frontend Code Arena United States vs Chinaarena · 折线图 · 23 个数据点从图片估算Kimi Code-K2.7 Code vs Kimi-K2.6 : Performance vs TokensKimi (Moonshot AI) · 散点图 · 6 个数据点从图片估算Agent Arena Top 25 Models - Kimi K3: Rank #4arena · 跑分柱状图 · 30 个数据点从图片估算相似重绘 7.1Artificial Analysis Intelligence Index v4.1artificial-analysis · 柱状图 · 44 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 跑分柱状图 · 72 个数据点从图片估算Kimi K2 Thinking vs GPT-5 vs Claude Sonnet 4.5 (Thinking) Benchmark ComparisonKimi (Moonshot AI) · 跑分柱状图 · 18 个数据点从图片估算MathVision Pass@1 vs Activated ParametersKimi (Moonshot AI) · 散点图 · 11 个数据点从图片估算Top ModelsKimi (Moonshot AI) · 堆叠柱状图 · 10 个数据点从图片估算t-SNE of MCP tools by CategoryKimi (Moonshot AI) · 散点图 · 34 个数据点从图片估算Time series dataKimi (Moonshot AI) · 折线图 · 20 个数据点从图片估算Proportion vs Stated ConfidenceKimi (Moonshot AI) · 折线图 · 125 个数据点从图片估算LLM Inference on M3 Max - Performance EvolutionKimi (Moonshot AI) · 散点图 · 9 个数据点从图片估算Per-step training loss curve of Kimi K2Kimi (Moonshot AI) · 折线图 · 18 个数据点从图片估算Figure 1: Kimi K2 main resultsKimi (Moonshot AI) · 跑分柱状图 · 56 个数据点从图片估算Kimi Code BenchKimi (Moonshot AI) · 柱状图 · 3 个数据点从图片估算Kimi K2.5 API PriceKimi (Moonshot AI) · 表格图 · 9 个数据点从图片估算LLM Inference on M3 Max - Performance EvolutionKimi (Moonshot AI) · 散点图 · 45 个数据点从图片估算τ²-Bench Telecom (Agentic Tool Use)Kimi (Moonshot AI) · 柱状图 · 25 个数据点从图片估算Kimi K2 Thinking vs GPT-5 vs Claude Sonnet 4.5 (Thinking)Kimi (Moonshot AI) · 跑分柱状图 · 18 个数据点从图片估算Output SpeedKimi (Moonshot AI) · 柱状图 · 2 个数据点从图片估算Kimi Code BenchKimi (Moonshot AI) · 柱状图 · 2 个数据点从图片估算Kimi K2.6 Agent via kimi.com's harnesses vs. Gemini 3.1 Pro via Google AI StudioKimi (Moonshot AI) · 柱状图 · 3 个数据点从图片估算General AgentsKimi (Moonshot AI) · 跑分柱状图 · 39 个数据点从图片估算Compare Categories by usecase on OpenRouterKimi (Moonshot AI) · 堆叠柱状图 · 10 个数据点从图片估算AI Office Bench vs General-Agent BenchKimi (Moonshot AI) · 跑分柱状图 · 6 个数据点从图片估算Figure 2: Max Logits vs Training StepsKimi (Moonshot AI) · 折线图 · 20 个数据点从图片估算MMMU ProKimi (Moonshot AI) · 折线图 · 100 个数据点从图片估算Execution Time to Solve Tasks of Different ComplexityKimi (Moonshot AI) · 散点图 · 34 个数据点从图片估算Multi-Objective Performance OptimizationKimi (Moonshot AI) · 散点图 · 10 个数据点从图片估算Net worth over timeKimi (Moonshot AI) · 折线图 · 45 个数据点从图片估算AI Model Performance ComparisonKimi (Moonshot AI) · 柱状图 · 14 个数据点从图片估算CodingKimi (Moonshot AI) · 跑分柱状图 · 41 个数据点从图片估算Model Comparison: Cost vs PerformanceKimi (Moonshot AI) · 折线图 · 6 个数据点从图片估算Kimi Claw BenchKimi (Moonshot AI) · 分组柱状图 · 2 个数据点从图片估算Kimi K2.5 Agent Swarm vs Kimi K2.5 vs Claude Opus 4.5Kimi (Moonshot AI) · 跑分柱状图 · 9 个数据点从图片估算Model Performance Across BenchmarksKimi (Moonshot AI) · 跑分柱状图 · 48 个数据点从图片估算Training Accuracy vs StepsKimi (Moonshot AI) · 折线图 · 100 个数据点从图片估算OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer EnvironmentsKimi (Moonshot AI) · 表格图 · 5 个数据点从图片估算Harvey LAB-AA against hallucinations: Criterion Pass Rate vs. Material Hallucinations per Taskartificial-analysis · 散点图 · 29 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 跑分柱状图 · 101 个数据点从图片估算Output Tokens Used to Run Artificial Analysis Intelligence Index: Non-reasoning Modelsartificial-analysis · 柱状图 · 17 个数据点从图片估算Kimi K3 2.8T Agentic Revenue & Profit Estimatessemianalysis · 堆叠柱状图 · 16 个数据点从图片估算Few flagship releases by China's leading AI labs are open under a restrictive licenseepoch-ai · 表格图 · 71 个数据点从图片估算AA-Briefcase Elo - Cost per Task (USD)artificial-analysis · 散点图 · 38 个数据点从图片估算Code Arena: WebDevarena · 柱状图 · 15 个数据点从图片估算Cost (USD) to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 40 个数据点从图片估算Artificial Analysis Intelligence Index v3.0 incorporates 10 evaluations: MMLU-Pro, GPQA Diamond, Humanity's Last Exam, LiveCodeBench, SciCode, AIME 2025, IFBench, AA-LCR, Terminal-Bench Hard, τ²-Bench Telecomartificial-analysis · 跑分柱状图 · 35 个数据点从图片估算Vision Kimi-K2.6: Top Open Modelarena · 柱状图 · 15 个数据点从图片估算Weighted cost per Intelligence Index Taskartificial-analysis · 跑分柱状图 · 88 个数据点从图片估算Cost per GPU/Total Tokens (Owning - Hyperscaler) vs. Interactivitysemianalysis · 散点图 · 36 个数据点从图片估算Text Kimi-K3: Overall Ranked #9arena · 柱状图 · 23 个数据点从图片估算Token Throughput per GPU vs. Interactivitysemianalysis · 折线图 · 18 个数据点从图片估算Cost (USD) to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 78 个数据点从图片估算Cost per Intelligence Indexartificial-analysis · 跑分柱状图 · 114 个数据点从图片估算Get Data DescriptionKimi (Moonshot AI) · 表格图 · 4 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 柱状图 · 26 个数据点从图片估算Artificial Analysis Intelligence Index v3.0artificial-analysis · 柱状图 · 33 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 柱状图 · 32 个数据点从图片估算Figure 1: Performance of Kimi-Audio and previous audio langauge models on various benchmarksKimi (Moonshot AI) · 表格图 · 75 个数据点从图片估算Kimi K2.5 Instant #2 open model in Vision Arenaarena · 表格图 · 11 个数据点从图片估算Output Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 58 个数据点从图片估算Qwen3.5-397B-A17B #top2 overall model in Visionarena · 表格图 · 28 个数据点从图片估算Agent Kimi K2.7 Code Leaderboard: Ranked #19arena · 跑分柱状图 · 25 个数据点从图片估算Output Tokens Used to Run All Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 44 个数据点从图片估算Token Efficiency before and after Toggle across BenchmarksKimi (Moonshot AI) · 表格图 · 14 个数据点从图片估算LMArena.ai WebDev Arena Leaderboardarena · 表格图 · 33 个数据点从图片估算Output tokens to run all evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 87 个数据点从图片估算Cost per Intelligence Index Taskartificial-analysis · 散点图 · 31 个数据点从图片估算Performance comparison across different vision-text joint-training strategiesKimi (Moonshot AI) · 表格图 · 30 个数据点从图片估算Model Performance ComparisonKimi (Moonshot AI) · 跑分柱状图 · 25 个数据点从图片估算Pass Rate on MiniF2F-testKimi (Moonshot AI) · 折线图 · 57 个数据点从图片估算PerceptionBench (Ours)Kimi (Moonshot AI) · 表格图 · 12 个数据点从图片估算Evaluation ResultsKimi (Moonshot AI) · 表格图 · 58 个数据点从图片估算Evaluation Time: 2025-10-10Kimi (Moonshot AI) · 表格图 · 107 个数据点从图片估算The AI that actually does thingsKimi (Moonshot AI) · 堆叠柱状图 · 32 个数据点从图片估算VoxelBench LeaderboardKimi (Moonshot AI) · 表格图 · 81 个数据点
Z.ai 78
全部模型发布方的数字相似重绘 9.0LLM Performance EvaluationZ.ai · 跑分柱状图 · 30 个数据点部分为发布方的数字LLM Benchmarks: Terminal Bench 2.1, DeepSWE v1.1, Agents' Last Exam, AutomationBench v1.0.6, HLE w/ Tools, GDPVal-AA v2Z.ai · 跑分柱状图 · 33 个数据点从图片估算相似重绘 7.7Coding Performance SWE-Bench Pro EvaluationZ.ai · 柱状图 · 7 个数据点从图片估算相似重绘 7.9Coding Performance EvaluationZ.ai · 柱状图 · 7 个数据点部分为发布方的数字相似重绘 9.2LLM Performance Bench 3.0Z.ai · 跑分柱状图 · 47 个数据点部分为发布方的数字8 LLM benchmarks: Humanity's Last Exam, SWE-bench Verified, SWE-bench Multilingual, Terminal-Bench 2.0, BrowseComp, MCP-Atlas, τ²-Bench, Vending Bench 2Z.ai · 跑分柱状图 · 57 个数据点部分为发布方的数字相似重绘 8.1Model Performance ComparisonZ.ai · 跑分柱状图 · 35 个数据点部分为发布方的数字8 LLM benchmarks Performance: AIME 25, LiveCodeBench v6, GPQA, HLE, SWE-bench Verified, Terminal-Bench, τ²-Bench, BrowseCompZ.ai · 跑分柱状图 · 40 个数据点从图片估算Code Arena 3.7 Max (20260517) : #3arena · 柱状图 · 19 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 表格图 · 33 个数据点从图片估算相似重绘 7.6Artificial Analysis Intelligence Index v4.1artificial-analysis · 柱状图 · 41 个数据点从图片估算相似重绘 8.3Code Arena: Frontend #2arena · 柱状图 · 14 个数据点从图片估算相似重绘 7.7Agent Arena Leaderboardarena · 柱状图 · 27 个数据点从图片估算相似重绘 4.8Agentic Coding Performance by Effort LevelZ.ai · 折线图 · 11 个数据点从图片估算函数绘图板Z.ai · 折线图 · 64 个数据点从图片估算Model Performance ComparisonZ.ai · 跑分柱状图 · 10 个数据点从图片估算Open-source LLMs BenchmarksZ.ai · 表格图 · 68 个数据点从图片估算相似重绘 7.9Average Token Usage per InteractionZ.ai · 柱状图 · 4 个数据点部分为发布方的数字Long-Horizon Task EvaluationZ.ai · 跑分柱状图 · 15 个数据点从图片估算Z.ai Agentic Code Bench v1.0, evaluated on Claude Code 2.1.207Z.ai · 折线图 · 15 个数据点从图片估算Model PerformanceZ.ai · 柱状图 · 10 个数据点从图片估算Optimization Iteration vs QPSZ.ai · 散点图 · 7 个数据点从图片估算相似重绘 8.58 LLM benchmarks: AIME 25, GPQA, LiveCodeBench v6, HLE, BrowseComp, SWE-bench Verified, Terminal-Bench, τ²-BenchZ.ai · 跑分柱状图 · 44 个数据点从图片估算LLM Performance on Harvey LABZ.ai · 柱状图 · 7 个数据点从图片估算CC-Bench: GLM-4.5's Experience with Agentic Coding in Real-world Development ScenariosZ.ai · 堆叠柱状图 · 24 个数据点从图片估算Artificial Analysis Intelligence Index v3.0Z.ai · 柱状图 · 44 个数据点从图片估算相似重绘 7.8LLM Benchmarks: SWE-bench Verified, τ²-Bench, BrowseComp, AIME 25, GPQA, HLEZ.ai · 跑分柱状图 · 18 个数据点从图片估算Average Tool Usage per InteractionZ.ai · 柱状图 · 4 个数据点从图片估算CC-Benchmark-V1.1: GLM 4.6s Experience with Agentic Coding in Real-worldZ.ai · 跑分柱状图 · 28 个数据点从图片估算Money Balance Over TimeZ.ai · 折线图 · 56 个数据点从图片估算CC-Bench-V2: GLM-4.7 vs. GLM-5 vs. Claude Opus 4.5Z.ai · 跑分柱状图 · 15 个数据点从图片估算Frontier Models: Intelligence vs. Cost per TaskZ.ai · 散点图 · 6 个数据点从图片估算相似重绘 7.0Coding EvaluationZ.ai · 柱状图 · 3 个数据点部分为发布方的数字LLM Benchmarks: Humanity's Last Exam, SWE-bench Verified, SWE-bench Multilingual, Terminal-Bench 2.0, BrowseComp, MCP-Atlas, τ²-Bench, Vending Bench 2Z.ai · 跑分柱状图 · 51 个数据点部分为发布方的数字8 LLM Benchmarks: SWE-bench Pro, Terminal-Bench 2.1 (Terminus), NL2Repo, DeepSWE, ProgramBench, MCP-Atlas, Tool-Decathlon, Humanity's Last ExamZ.ai · 跑分柱状图 · 52 个数据点从图片估算AutoGLM Success Rates on Frequent TasksZ.ai · 堆叠柱状图 · 21 个数据点从图片估算GLM-4.7 vs GLM-4.6 in Real-world Development ScenariosZ.ai · 堆叠柱状图 · 9 个数据点从图片估算Model Parameters vs SWE-bench VerifiedZ.ai · 散点图 · 9 个数据点从图片估算LLM Performance Evaluation: Agentic, Reasoning, And Coding BenchmarksZ.ai · 柱状图 · 14 个数据点从图片估算Agentic Coding: SWE-Bench ProZ.ai · 柱状图 · 7 个数据点从图片估算GLM-4.5: Agentic, Reasoning, and Coding (ARC)Z.ai · 跑分柱状图 · 56 个数据点从图片估算相似重绘 9.28 LLM benchmarks: AIME 25, GPQA, LiveCodeBench v6, HLE, BrowseComp, SWE-bench Verified, Terminal-Bench, τ²-BenchZ.ai · 跑分柱状图 · 43 个数据点从图片估算KernelBench Level 3: Agentic ML Workload OptimizationZ.ai · 折线图 · 4 个数据点从图片估算Model Comparison MetricsZ.ai · 表格图 · 27 个数据点从图片估算GLM-4.5's Experience with Agentic Coding in Real-world Development ScenariosZ.ai · 堆叠柱状图 · 9 个数据点从图片估算Accuracy ComparisonZ.ai · 柱状图 · 4 个数据点从图片估算BenchmarksZ.ai · 表格图 · 36 个数据点从图片估算Open-source LLMs BenchmarksZ.ai · 表格图 · 39 个数据点从图片估算CC-Bench-V1.1: GLM 4.6's Experience with Agentic Coding in Real-world Development ScenariosZ.ai · 堆叠柱状图 · 12 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 跑分柱状图 · 50 个数据点从图片估算AGENTIC WEBDEVarena · 表格图 · 10 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 37 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 柱状图 · 35 个数据点从图片估算Output Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 33 个数据点从图片估算Elo rating-AA v2.1 on real-world work tasksartificial-analysis · 柱状图 · 28 个数据点从图片估算Cost per Intelligence Index Taskartificial-analysis · 柱状图 · 25 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 分组柱状图 · 27 个数据点从图片估算Artificial Analysis Cyber Indexartificial-analysis · 跑分柱状图 · 62 个数据点从图片估算GLM-5.2 sets a new ECI record for open-weight models, but remains far behind state-of-the-art performanceepoch-ai · 跑分柱状图 · 14 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 分组柱状图 · 27 个数据点从图片估算Text Arena GLM-5.2 (Max) vs GLM-5.1 Category Rankingsarena · 表格图 · 28 个数据点从图片估算Intelligence vs. Cost per Intelligence Index Taskartificial-analysis · 散点图 · 33 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 跑分柱状图 · 41 个数据点从图片估算Output Tokens used to run evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 堆叠柱状图 · 82 个数据点从图片估算Third parties took over serving GLM 5.3 Flash within three weeks of its open-weight releaseepoch-ai · 堆叠柱状图 · 4 个数据点从图片估算Artificial Analysis Intelligence Indexartificial-analysis · 分组柱状图 · 44 个数据点从图片估算GLM-5.2 (Max): Ranked #10arena · 表格图 · 12 个数据点从图片估算Code Arena: WebDev Rank Top 15arena · 柱状图 · 20 个数据点从图片估算Output Tokens used to run all evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 54 个数据点从图片估算Speed vs. Price: GLM-4.7 Providersartificial-analysis · 散点图 · 8 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 柱状图 · 36 个数据点从图片估算GDPval-AA scores for agentic performance on real-world work tasks using web and shell access via Stirrup, an open-source harness developed by Artificial Analysisartificial-analysis · 柱状图 · 30 个数据点从图片估算Output Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 36 个数据点从图片估算Intelligence vs. Cost to Run Artificial Analysis Indexartificial-analysis · 散点图 · 20 个数据点从图片估算Code Arena: Frontend GLM-5.2 (Max): Head-to-Headarena · 跑分柱状图 · 42 个数据点从图片估算Pareto Frontier for Code Arena: WebDevarena · 散点图 · 17 个数据点从图片估算AA-Omniscence Indexartificial-analysis · 分组柱状图 · 32 个数据点从图片估算Per-Layer KV-Cache SizeZ.ai · 折线图 · 24 个数据点
Apple 33
全部模型从图片估算相似重绘 9.0GPQA Diamond (Scientific Reasoning)artificial-analysis · 柱状图 · 19 个数据点从图片估算System Output Throughput vs Output Speed per Queryartificial-analysis · 折线图 · 95 个数据点从图片估算Accuracy / Percentile (%) by Benchmarkdeepseek · 跑分柱状图 · 30 个数据点从图片估算Artificial Analysis Intelligence Index v2.2artificial-analysis · 柱状图 · 22 个数据点从图片估算Artificial Analysis Omniscence Index vs. Total Parametersartificial-analysis · 散点图 · 44 个数据点从图片估算Performance vs Latency for MobileCLIP and Baselinesapple · 散点图 · 33 个数据点从图片估算deepseek-r1-lite-preview AIME accuracy with test-time compute scalingdeepseek · 折线图 · 17 个数据点从图片估算Model Accuracy vs Latency Comparisonapple · 散点图 · 27 个数据点从图片估算Average Score (16K max context) vs. End-to-End Generation Timeartificial-analysis · 散点图 · 39 个数据点从图片估算GPQA Diamond vs. Inference Time; GPQA Diamond scoreartificial-analysis · 散点图 · 15 个数据点从图片估算Artificial Analysis Intelligence Index incorporates 7 evaluations: MMLU-Pro, GPQA Diamond, Humanity's Last Exam, LiveCodeBench, SciCode, AIME, MATH-500artificial-analysis · 跑分柱状图 · 21 个数据点从图片估算Agent MiniMax M3: Ranked #18arena · 柱状图 · 26 个数据点从图片估算Artificial Analysis Intelligence Index incorporates 7 evaluations: MMLU-Pro, GPQA Diamond, Humanity's Last Exam, LiveCodeBench, SciCode, MATH-500artificial-analysis · 柱状图 · 22 个数据点从图片估算MATH Level 5 accuracy for DeepSeek-R1-0528 compared to select leading modelsepoch-ai · 跑分柱状图 · 6 个数据点从图片估算Intelligence Parameters vs. Active Parametersartificial-analysis · 散点图 · 15 个数据点从图片估算Artificial Analysis Intelligence Index v2artificial-analysis · 柱状图 · 22 个数据点从图片估算Input/Output Pricing for o1-Class Inference Models ($/1M Tokens)deepseek · 分组柱状图 · 12 个数据点从图片估算SWE-bench verified accuracy for DeepSeek-R1-0528 compared to select leading modelsepoch-ai · 跑分柱状图 · 15 个数据点从图片估算Peak Inference Memoryartificial-analysis · 柱状图 · 32 个数据点从图片估算Artificial Analysis Intelligence Index incorporates 7 evaluations: MMLU-Pro, GPQA Diamond, Humanity's Last Exam, LiveCodeBench, SciCode, AIME, MATH-500artificial-analysis · 柱状图 · 26 个数据点从图片估算Intelligence vs. Output Tokens Used in Artificial Analysis Intelligence Indexartificial-analysis · 散点图 · 4 个数据点从图片估算Code MiniMax M3: Frontend Ranks #7arena · 柱状图 · 21 个数据点从图片估算Expert Prompts vs Open Frontier Gaparena · 分组柱状图 · 30 个数据点从图片估算Output Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 35 个数据点从图片估算Throughput per User vs Output Throughput per GPUsemianalysis · 折线图 · 30 个数据点从图片估算DeepSeek R1 O528 • FP8 • 8K / 1K • Source: SemiAnalysis InferenceX™ • Updated: 02/17/2026semianalysis · 散点图 · 19 个数据点从图片估算Health Age Historyapple · 折线图 · 5 个数据点从图片估算TCO $/GPU/Million Total Tokens (Owning) vs Interactivitysemianalysis · 折线图 · 12 个数据点从图片估算Model Performance vs Latencyapple · 散点图 · 13 个数据点从图片估算Interpolated Token Throughput per GPU (tok/s/gpu)semianalysis · 折线图 · 85 个数据点从图片估算Artificial Analysis Intelligence Index incorporates 7 evaluations: MMLU-Pro, GPQA Diamond, Humanity's Last Exam, LiveCodeBench, SciCode, AIME, MATH-500artificial-analysis · 柱状图 · 17 个数据点从图片估算AIME 2024deepseek · 柱状图 · 8 个数据点从图片估算Artificial Analysis Intelligence Index v4.0artificial-analysis · 柱状图 · 32 个数据点
MiniMax 18
从图片估算Model Performance ComparisonMiniMax · 跑分柱状图 · 18 个数据点部分为发布方的数字相似重绘 8.8Model Performance Comparison Across BenchmarksMiniMax · 跑分柱状图 · 45 个数据点部分为发布方的数字相似重绘 8.1Model Benchmarks ComparisonMiniMax · 跑分柱状图 · 36 个数据点部分为发布方的数字相似重绘 8.3Model Performance Comparison Across BenchmarksMiniMax · 跑分柱状图 · 46 个数据点从图片估算相似重绘 9.0Model Performance ComparisonMiniMax · 跑分柱状图 · 18 个数据点部分为发布方的数字Model Performance Comparison Across BenchmarksMiniMax · 跑分柱状图 · 63 个数据点从图片估算相似重绘 7.7Model Performance by SubsetMiniMax · 跑分柱状图 · 20 个数据点从图片估算Model Performance Comparison Across BenchmarksMiniMax · 跑分柱状图 · 33 个数据点部分为发布方的数字Model Benchmarks ComparisonMiniMax · 跑分柱状图 · 40 个数据点从图片估算相似重绘 7.8Model Performance ComparisonMiniMax · 跑分柱状图 · 28 个数据点从图片估算SWE-bench Verified Score EvolutionMiniMax · 折线图 · 33 个数据点从图片估算相似重绘 8.3Medal Rates Over TimeMiniMax · 折线图 · 125 个数据点从图片估算ELO scores-AA gold subset of GDPval, independently assessed by Artificial Analysisartificial-analysis · 柱状图 · 32 个数据点从图片估算Artificial Analysis Intelligence Index by Open Weights / Proprietaryartificial-analysis · 堆叠柱状图 · 47 个数据点发布方的数字相似重绘 6.4Model Benchmarks ComparisonMiniMax · 跑分柱状图 · 61 个数据点从图片估算Model Performance Across BenchmarksMiniMax · 跑分柱状图 · 50 个数据点从图片估算Model Performance Comparison Across SubsetsMiniMax · 跑分柱状图 · 20 个数据点从图片估算Cost (USD) to Run all Artificial Analysis Artificial Analysis Intelligence Indexartificial-analysis · 堆叠柱状图 · 29 个数据点
Mistral AI 31
从图片估算Model Performance Comparison (Instruct)Mistral AI · 跑分柱状图 · 8 个数据点从图片估算Base Model Benchmark ComparisonMistral AI · 跑分柱状图 · 20 个数据点从图片估算Score vs. Output LengthMistral AI · 跑分柱状图 · 20 个数据点从图片估算相似重绘 8.3Agentic Benchmarks vs previous Mistral coding modelsMistral AI · 柱状图 · 4 个数据点从图片估算Math, instruction following vs competing models*Mistral AI · 分组柱状图 · 37 个数据点从图片估算Agentic Benchmarks vs previous Mistral modelsMistral AI · 跑分柱状图 · 38 个数据点从图片估算相似重绘 8.1Agentic Benchmarks vs competing models*Mistral AI · 跑分柱状图 · 50 个数据点从图片估算Performance comparison across internal modelsMistral AI · 跑分柱状图 · 50 个数据点从图片估算Performance Comparison Across Internal ModelsMistral AI · 跑分柱状图 · 24 个数据点从图片估算Score vs. Output LengthMistral AI · 跑分柱状图 · 16 个数据点从图片估算Score vs. Output LengthMistral AI · 跑分柱状图 · 10 个数据点从图片估算Cost per Intelligence Index Taskartificial-analysis · 散点图 · 30 个数据点从图片估算Mistral Large 3 is a new ranking #28 in Text model overallarena · 表格图 · 18 个数据点从图片估算Output Reasoning Models Used to Run Artificial Analysis Intelligence Indexartificial-analysis · 柱状图 · 17 个数据点从图片估算Mistral Large 3 is the #1 open model in Codingarena · 表格图 · 36 个数据点从图片估算Vals.ai - Harvey's Legal Agent BenchmarkMistral AI · 柱状图 · 8 个数据点从图片估算Figure 1: Confidence Intervals on Model Strength (via Bootstrapping)arena · 柱状图 · 25 个数据点从图片估算Pricing per 1M Tokens; Lower and Output is better Pricesartificial-analysis · 分组柱状图 · 6 个数据点从图片估算Mistral Medium 3 ranks at #11 in the Arena!arena · 表格图 · 52 个数据点从图片估算Model Accuracy Comparisonartificial-analysis · 柱状图 · 6 个数据点从图片估算Pricing: USD per 1M Tokensartificial-analysis · 分组柱状图 · 28 个数据点从图片估算Intelligence vs. Active Parametersartificial-analysis · 散点图 · 30 个数据点从图片估算LMArena ELO ScoreMistral AI · 柱状图 · 6 个数据点从图片估算Performance / MMLU-Proarena · 散点图 · 4 个数据点从图片估算Intelligence Evaluationsartificial-analysis · 柱状图 · 101 个数据点从图片估算Output Tokens used to Run all Evaluations in the Artificial Analysis Intelligence Indexartificial-analysis · 跑分柱状图 · 33 个数据点从图片估算Open Weights: Frontier Indexartificial-analysis · 折线图 · 45 个数据点从图片估算Artificial Analysis Intelligence Index v3.0artificial-analysis · 柱状图 · 25 个数据点从图片估算Artificial Analysis - AutomationBenchMistral AI · 柱状图 · 8 个数据点从图片估算Artificial Analysis Cyber IndexMistral AI · 柱状图 · 19 个数据点从图片估算MMLU vs. Priceartificial-analysis · 散点图 · 13 个数据点
常见问题
什么是图表数据？
对实验室或榜单发布的每张图，我们提供一个文件，包含图上的每个数字、数字的来源，以及我们有多确定。目前共 2391 张图、73,917 个数据点。
数字是精确的吗？
只有 2 张图完全来自发布方的文字数字，29 张部分来自。其余是从图片读取的，并标注为估算。每个页面都写明是哪一种。
可以下载文件吗？
可以。每张图的页面都有 JSON、CSV 和 agent skill 文件，在邮箱解锁区后面。API 页面列出了静态文件。