🏆 AI Benchmarks Index

Graduate-level science questions designed to be Google-proof — tests deep reasoning.

1. 1Gemini 4 Argongoogle96.4%
2. 2GPT-6 Astraopenai95.8%
3. 3Claude Sonnet 5.5anthropic95.6%
4. 4Gemini 3.8 Flashgoogle95.4%
5. 5GPT-6.1 Solopenai95.4%
6. 6Gemini 3.7 Flashgoogle94.8%
7. 7GPT-5.4 Proopenai94.6%
8. 8GPT-6 Solopenai94.3%
9. 9Gemini 3.6 Flashgoogle94.1%
10. 10GPT-5.5openai94.0%

Real GitHub issues an agent must fix end-to-end — the standard for coding agents.

1. 1Claude Opus 4.7 (Fast)anthropic83.5%
2. 2Claude Opus 4.7anthropic83.5%
3. 3GPT-5.5openai80.6%
4. 4Gemini 3.5 Flashgoogle79.3%
5. 5Claude Opus 4.6anthropic78.7%
6. 6GLM 5.2z-ai78.7%
7. 7DeepSeek V4 Prodeepseek77.6%
8. 8Qwen3.7 Maxqwen77.3%
9. 9GPT-5.4openai76.9%
10. 10Kimi K2.6moonshotai76.7%

The hardest tier of competition math problems from the MATH benchmark.

1. 1GPT-5openai98.1%
2. 2GPT-5 Miniopenai97.8%
3. 3o4 Miniopenai97.8%
4. 4o3openai97.8%
5. 5Claude Sonnet 4.5anthropic97.7%
6. 6Qwen3 Maxqwen97.1%
7. 7R1deepseek96.6%
8. 8o3 Miniopenai96.5%
9. 9Claude Haiku 4.5anthropic96.4%
10. 10Gemini 2.5 Progoogle95.9%

Olympiad-style mock AIME exam (2024–25) — elite high-school competition math.

1. 1GPT-6 Solopenai100.0%
2. 2Claude Opus 5.5anthropic100.0%
3. 3GPT-6 Astraopenai100.0%
4. 4GPT-5.5openai100.0%
5. 5Claude Fable 5.1anthropic100.0%
6. 6GPT-5.5 Proopenai100.0%
7. 7Qwen3.8 Max (0902)qwen100.0%
8. 8GPT-5.6 Solopenai100.0%
9. 9Claude Fable 5anthropic100.0%
10. 10GPT-6.1 Solopenai100.0%

Research-level math problems written by mathematicians — the hardest math benchmark.

1. 1GPT-5.5 Proopenai52.4%
2. 2GPT-5.5openai51.7%
3. 3GPT-5.4 Proopenai50.0%
4. 4GPT-5.4openai47.6%
5. 5Claude Opus 4.8 (Fast)anthropic47.2%
6. 6Claude Opus 4.8anthropic47.2%
7. 7Claude Opus 4.7 (Fast)anthropic43.8%
8. 8Claude Opus 4.7anthropic43.8%
9. 9GPT-5.2openai40.7%
10. 10Claude Opus 4.6anthropic40.7%

Short factual questions measuring hallucination resistance — can it just be right?

1. 1Gemini 4 Argongoogle76.0%
2. 2GPT-6 Astraopenai75.6%
3. 3GPT-6.1 Solopenai73.9%
4. 4Claude Opus 5.5anthropic72.2%
5. 5Claude Fable 5.1anthropic70.8%
6. 6Claude Fable 5anthropic70.7%
7. 7Gemini 3.8 Flashgoogle69.7%
8. 8GPT-5.6 Solopenai69.7%
9. 9Gemini 3.7 Flashgoogle69.2%
10. 10Gemini 3.6 Flashgoogle66.2%

Human preference Elo from blind head-to-head chat battles — the de-facto vibes benchmark.

1. 1claude-opus-5-highanthropic1503
2. 2Claude Opus 4.6anthropic1498
3. 3Claude Fable 5anthropic1493
4. 4claude-fable-5-highanthropic1491
5. 5MiMo-V2.6-Proxiaomi1491
6. 6Qwen3.8 Maxqwen1483
7. 7Claude Opus 4.7anthropic1483
8. 8Gemini 3.5 Flashgoogle1482
9. 9Gemini 3.1 Pro Previewgoogle1480
10. 10Gemini 3.6 Flashgoogle1479

Human preference Elo for text-to-image generation battles.

1. 1gpt-image-2.5-sunburstopenai1425
2. 2gpt-image-2.5-flareopenai1398
3. 3gpt-image-2 (medium)openai1383
4. 4grok-imagine-image-2.0 (canvas)xai1334
5. 5mai-image-2.6microsoft-ai1332
6. 6Nano Banana 2.1google1328
7. 7reve-2.1reve1302
8. 8grok-imagine-image-2.0 (low)xai1301
9. 9grok-imagine-image-2.0 (20260801)xai1294
10. 10muse-imagemeta1274

Human preference Elo for image-to-video generation battles.

1. 1minimax-h3minimax1495
2. 2gemini-omni-1.1-flashgoogle1489
3. 3wan3.0alibaba1480
4. 4dreamina-seedance-2.5-720pbytedance1477
5. 5dreamina-seedance-2.0-720pbytedance1475
6. 6grok-imagine-video-1.5-preview-720pxai1466
7. 7gemini-omni-flashgoogle1465
8. 8hidream-o1-video-1.0hidream1457
9. 9grok-imagine-video-1.5-720pxai1456
10. 10flux-3-video-20260811bfl1449