- 图表数据
- artificial-analysis
- Intelligence Evaluations
artificial-analysis跑分柱状图 · 275 个数据点 · Muse Spark
Intelligence Evaluations
artificial-analysis 发布的一张图背后的数字，用原生图表重画。对照原图，研究数据，再做你自己的。
简化重绘
由我们用下面的数字重新绘制。悬停、点按或用方向键查看。这是简化重绘：它用我们自己的样式展示数字，不是原图的样子。强调色标出这张图的主角。表格里带 ~ 的是估算值。
数字
275 个点中前 10 个免费| 分组 | 系列 | 数值 | 来源 | 置信 |
|---|---|---|---|---|
| OpenAI | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| Anthropic | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 | |
| Alibaba | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| xAI | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| Meta | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| DeepSeek | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| Mistral | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| NVIDIA | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| Amazon | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
Max 数据，目前留邮箱免费看
查看全部 275 个数据点并下载文件
这张图的完整表格、JSON、CSV 和 agent skill 文件。它们属于 Max；原型阶段，留邮箱即可解锁。
| Zhipu AI | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
|---|---|---|---|---|
| Moonshot AI | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| 01.AI | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| Baidu | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| Tencent | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| AI2 | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| ServiceNow | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| Cohere | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| IBM | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| LG AI Research | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| Microsoft | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| Samsung | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| FPT Software | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| Tsinghua | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| Others | GDPval-AA (Agnetic Real-World Tasks, (ELO-500)/2000) | ~100% | 估算 | 0.50 |
| OpenAI | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~54% | 估算 | 0.50 |
| Anthropic | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~53% | 估算 | 0.50 |
| Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~52% | 估算 | 0.50 | |
| Alibaba | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| xAI | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| Meta | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| DeepSeek | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| Mistral | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| NVIDIA | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| Amazon | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| Zhipu AI | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| Moonshot AI | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| 01.AI | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| Baidu | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| Tencent | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| AI2 | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| ServiceNow | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| Cohere | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| IBM | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| LG AI Research | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| Microsoft | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| Samsung | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| FPT Software | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| Tsinghua | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| Others | Terminal-Bench Hard (Agnetic Coding & Terminal Use) | ~50% | 估算 | 0.50 |
| OpenAI | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Anthropic | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 | |
| Alibaba | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| xAI | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Meta | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| DeepSeek | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Mistral | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| NVIDIA | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Amazon | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Zhipu AI | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Moonshot AI | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| 01.AI | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Baidu | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Tencent | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| AI2 | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| ServiceNow | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Cohere | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| IBM | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| LG AI Research | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Microsoft | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Samsung | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| FPT Software | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Tsinghua | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| Others | 1st-Bench Telecom (Agnetic Tool Use) | ~96% | 估算 | 0.50 |
| OpenAI | AA-LCR (Long Context Reasoning) | ~74% | 估算 | 0.50 |
| Anthropic | AA-LCR (Long Context Reasoning) | ~73% | 估算 | 0.50 |
| AA-LCR (Long Context Reasoning) | ~71% | 估算 | 0.50 | |
| Alibaba | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| xAI | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| Meta | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| DeepSeek | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| Mistral | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| NVIDIA | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| Amazon | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| Zhipu AI | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| Moonshot AI | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| 01.AI | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| Baidu | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| Tencent | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| AI2 | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| ServiceNow | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| Cohere | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| IBM | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| LG AI Research | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| Microsoft | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| Samsung | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| FPT Software | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| Tsinghua | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| Others | AA-LCR (Long Context Reasoning) | ~70% | 估算 | 0.50 |
| OpenAI | AA-Omniscence Accuracy (Knowledge) | ~55% | 估算 | 0.50 |
| Anthropic | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 | |
| Alibaba | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| xAI | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| Meta | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| DeepSeek | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| Mistral | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| NVIDIA | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| Amazon | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| Zhipu AI | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| Moonshot AI | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| 01.AI | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| Baidu | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| Tencent | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| AI2 | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| ServiceNow | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| Cohere | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| IBM | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| LG AI Research | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| Microsoft | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| Samsung | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| FPT Software | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| Tsinghua | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| Others | AA-Omniscence Accuracy (Knowledge) | ~54% | 估算 | 0.50 |
| OpenAI | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~83% | 估算 | 0.50 |
| Anthropic | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 | |
| Alibaba | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| xAI | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| Meta | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| DeepSeek | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| Mistral | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| NVIDIA | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| Amazon | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| Zhipu AI | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| Moonshot AI | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| 01.AI | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| Baidu | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| Tencent | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| AI2 | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| ServiceNow | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| Cohere | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| IBM | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| LG AI Research | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| Microsoft | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| Samsung | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| FPT Software | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| Tsinghua | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| Others | AA-Omniscence Non-Hallucination Rate (1 - Hallucination Rate) | ~78% | 估算 | 0.50 |
| OpenAI | Humanity's Last Exam (Reasoning & Knowledge) | ~45% | 估算 | 0.50 |
| Anthropic | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 | |
| Alibaba | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| xAI | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Meta | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| DeepSeek | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Mistral | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| NVIDIA | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Amazon | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Zhipu AI | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Moonshot AI | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| 01.AI | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Baidu | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Tencent | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| AI2 | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| ServiceNow | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Cohere | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| IBM | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| LG AI Research | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Microsoft | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Samsung | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| FPT Software | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Tsinghua | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| Others | Humanity's Last Exam (Reasoning & Knowledge) | ~42% | 估算 | 0.50 |
| OpenAI | GPQA Diamond (Scientific Reasoning) | ~84% | 估算 | 0.50 |
| Anthropic | GPQA Diamond (Scientific Reasoning) | ~92% | 估算 | 0.50 |
| GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 | |
| Alibaba | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| xAI | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| Meta | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| DeepSeek | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| Mistral | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| NVIDIA | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| Amazon | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| Zhipu AI | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| Moonshot AI | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| 01.AI | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| Baidu | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| Tencent | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| AI2 | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| ServiceNow | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| Cohere | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| IBM | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| LG AI Research | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| Microsoft | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| Samsung | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| FPT Software | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| Tsinghua | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| Others | GPQA Diamond (Scientific Reasoning) | ~90% | 估算 | 0.50 |
| OpenAI | SciCode (Coding) | ~55% | 估算 | 0.50 |
| Anthropic | SciCode (Coding) | ~53% | 估算 | 0.50 |
| SciCode (Coding) | ~53% | 估算 | 0.50 | |
| Alibaba | SciCode (Coding) | ~53% | 估算 | 0.50 |
| xAI | SciCode (Coding) | ~53% | 估算 | 0.50 |
| Meta | SciCode (Coding) | ~53% | 估算 | 0.50 |
| DeepSeek | SciCode (Coding) | ~53% | 估算 | 0.50 |
| Mistral | SciCode (Coding) | ~53% | 估算 | 0.50 |
| NVIDIA | SciCode (Coding) | ~53% | 估算 | 0.50 |
| Amazon | SciCode (Coding) | ~53% | 估算 | 0.50 |
| Zhipu AI | SciCode (Coding) | ~53% | 估算 | 0.50 |
| Moonshot AI | SciCode (Coding) | ~53% | 估算 | 0.50 |
| 01.AI | SciCode (Coding) | ~53% | 估算 | 0.50 |
| Baidu | SciCode (Coding) | ~53% | 估算 | 0.50 |
| Tencent | SciCode (Coding) | ~53% | 估算 | 0.50 |
| AI2 | SciCode (Coding) | ~53% | 估算 | 0.50 |
| ServiceNow | SciCode (Coding) | ~53% | 估算 | 0.50 |
| Cohere | SciCode (Coding) | ~53% | 估算 | 0.50 |
| IBM | SciCode (Coding) | ~53% | 估算 | 0.50 |
| LG AI Research | SciCode (Coding) | ~53% | 估算 | 0.50 |
| Microsoft | SciCode (Coding) | ~53% | 估算 | 0.50 |
| Samsung | SciCode (Coding) | ~53% | 估算 | 0.50 |
| FPT Software | SciCode (Coding) | ~53% | 估算 | 0.50 |
| Tsinghua | SciCode (Coding) | ~53% | 估算 | 0.50 |
| Others | SciCode (Coding) | ~53% | 估算 | 0.50 |
| OpenAI | IFBench (Instruction Following) | ~81% | 估算 | 0.50 |
| Anthropic | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| IFBench (Instruction Following) | ~79% | 估算 | 0.50 | |
| Alibaba | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| xAI | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| Meta | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| DeepSeek | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| Mistral | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| NVIDIA | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| Amazon | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| Zhipu AI | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| Moonshot AI | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| 01.AI | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| Baidu | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| Tencent | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| AI2 | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| ServiceNow | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| Cohere | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| IBM | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| LG AI Research | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| Microsoft | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| Samsung | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| FPT Software | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| Tsinghua | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| Others | IFBench (Instruction Following) | ~79% | 估算 | 0.50 |
| OpenAI | CrIPI (Physics Reasoning) | ~30% | 估算 | 0.50 |
| Anthropic | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 | |
| Alibaba | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| xAI | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| Meta | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| DeepSeek | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| Mistral | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| NVIDIA | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| Amazon | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| Zhipu AI | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| Moonshot AI | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| 01.AI | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| Baidu | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| Tencent | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| AI2 | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| ServiceNow | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| Cohere | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| IBM | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| LG AI Research | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| Microsoft | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| Samsung | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| FPT Software | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| Tsinghua | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
| Others | CrIPI (Physics Reasoning) | ~27% | 估算 | 0.50 |
下载这张图的数据
Max 订阅者可以通过 API 取同样的数据。 API 文档
API · Max
用 API 调用这张图
Max 套餐可以取这张图的数字（每个数字都带来源），格式为 JSON 或 CSV，附重画规格；也可以让智能体通过 MCP 调用。图表 id：e3eccf066faf。
| 路径 | 返回 | 套餐 | |
|---|---|---|---|
| GET | /api/v1/charts/e3eccf066faf | 标题、发布者、日期、图型、数据点数；用 Max 密钥时附数据链接 | 开放 |
| GET | /api/v1/charts/e3eccf066faf/data | 每个数据点及其来源和置信度，加上重画规格。加 ?format=csv 得到 CSV | Max |
| GET | /api/v1/charts/e3eccf066faf/render?view=1 | 本页使用的只读绘图视图 | 开放 |
| POST | /mcp | 工具 get_chart_data 加这个 id，供 Claude、Cursor 等智能体使用 | Max |
curlMax
curl -s https://viral-chart.com/api/v1/charts/e3eccf066faf/data \
-H "Authorization: Bearer $VIRALCHART_KEY"Claude CodeMax
claude mcp add --transport http viralchart https://viral-chart.com/mcp \
--header "Authorization: Bearer $VIRALCHART_KEY"每个密钥每分钟 60 次，超出返回 429 和 Retry-After。每个账号最多 5 个密钥，在账户页创建，只显示一次。没有 Max 时数据接口返回 402，并附套餐链接。这些数字描述的是发布者拥有的图，每个文件都带授权说明。
做一张这样的图
这张图的形式，你的数字。
实时预览。所有内容都留在你的设备上。
选择形式
简化形式：用我们自己的样式，不是原图的样子。你的数字从这张图的数据开始。
换上你的数字
这是这张图自己的名称和数字，方便你看清形式。直接改写任何名称或数字，预览会随之重画。发布前请把每一个都换成你的。
这是这张图的数字，方便你看清形式。发布前请把每一个都换成你的。
你的品牌
强调色
Logo（可选）只在你的设备上绘制，不会上传。
画面
导出图片
免费图片的角落有一行小字 “Made with ViralChart”，别人就是通过这一行找到我们的。
Meta 的更多数据
全部图表数据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
常见问题
这些数字有多准？
它们由两遍独立读图估算得出，置信度 0.50（0 到 1）。图上印着的数字读得准，从坐标轴读出的值可能有几个百分点的偏差。
这些数据可以直接用吗？
这些数字描述的是 artificial-analysis 发布的图表。图片、版式和样式归他们所有。使用数字时请注明发布者，图片请用你自己的数据来做。
怎么拿到文件？
JSON、CSV 和 agent skill 文件在本页的邮箱解锁区。原型里它们是静态文件：/api/charts/e3eccf066faf.json 和 /api/charts/e3eccf066faf.skill.md。