Benchmarks · other
Arena Vision · Homework
Human preference on prompts with image inputs (multimodal understanding); Bradley-Terry rating. Only the Homework category.
87 models · 96 results · data as of 2026-10-29 · arena.ai ↗
Leaderboard
GPT-6.1 Sol max
All results
96 rows| # | Model | Effort | Arena Vision · Homework |
|---|---|---|---|
| 1 | GPT-6.1 Sol | max | 1358 ±31 |
| 2 | Fable 5 | high | 1352 ±15 |
| 3 | Opus 5.5 | high | 1351 ±29 |
| 4 | GPT-6 Astra | max | 1342 ±25 |
| 5 | Gemini 3.8 Flash | high | 1338 ±22 |
| 6 | GPT-5.5 | high | 1335 ±11 |
| 7 | Fable 5.1 | max | 1334 ±21 |
| 8 | Qwen 3.8 Max | max | 1334 ±19 |
| 9 | Opus 4.8 | high | 1332 ±14 |
| 10 | Opus 4.7 | high | 1331 ±12 |
| 11 | GPT-5.5 | xhigh | 1330 ±11 |
| 12 | GPT-5.4 | high | 1330 ±10 |
| 13 | Opus 4.7 | max | 1326 ±12 |
| 14 | Sonnet 5.5 | xhigh | 1326 ±30 |
| 15 | Opus 4.6 | high | 1325 ±12 |
| 16 | Grok 4.7 | xhigh | 1324 ±25 |
| 17 | Opus 5 | high | 1323 ±14 |
| 18 | GPT-5.6 Terra | xhigh | 1321 ±17 |
| 19 | Opus 4.8 | max | 1321 ±13 |
| 20 | Gemini 3.5 Flash | medium | 1316 ±15 |
| 21 | GPT-5.6 Sol | xhigh | 1314 ±17 |
| 22 | Opus 4.6 | max | 1312 ±11 |
| 23 | Gemini 3.1 Pro | high | 1310 ±8 |
| 24 | Gemini 3.7 Flash | high | 1309 ±23 |
| 25 | Gemini 3 Pro | high | 1309 ±14 |
| 26 | Gemini 3 Flash | high | 1308 ±9 |
| 27 | DeepSeek V4.1 Flash | max | 1307 ±25 |
| 28 | GLM 5.3 Flash | max | 1306 ±13 |
| 29 | GPT-6 Sol | max | 1306 ±26 |
| 30 | Gemini 3.5 Flash | high | 1306 ±15 |