Benchmarks · other
Arena Vision · Chinese
Human preference on prompts with image inputs (multimodal understanding); Bradley-Terry rating. Only the Chinese category.
90 models · 99 results · data as of 2026-10-29 · arena.ai ↗
Leaderboard
Gemini 3.8 Flash high
All results
99 rows| # | Model | Effort | Arena Vision · Chinese |
|---|---|---|---|
| 1 | Gemini 3.8 Flash | high | 1384 ±40 |
| 2 | Gemini 3.7 Flash | high | 1369 ±35 |
| 3 | Muse Spark 1.3 | xhigh | 1366 ±36 |
| 4 | Opus 5 | high | 1363 ±20 |
| 5 | GPT-6 Astra | max | 1358 ±37 |
| 6 | Fable 5 | high | 1355 ±23 |
| 7 | Opus 4.6 | high | 1352 ±18 |
| 8 | Fable 5.1 | max | 1351 ±35 |
| 9 | Qwen 3.8 Max | max | 1350 ±26 |
| 10 | Opus 4.6 | max | 1348 ±17 |
| 11 | Muse Spark 1.3 | max | 1347 ±35 |
| 12 | Opus 4.7 | max | 1347 ±18 |
| 13 | Gemini 3.5 Flash | high | 1346 ±22 |
| 14 | Gemini 3.5 Flash | medium | 1343 ±22 |
| 15 | Gemini 3 Pro | high | 1343 ±17 |
| 16 | Gemini 3.1 Pro | high | 1339 ±13 |
| 17 | GLM 5.3 Flash | max | 1338 ±23 |
| 18 | Opus 4.7 | high | 1336 ±18 |
| 19 | Muse Spark 1.1 | max | 1335 ±23 |
| 20 | Gemini 3.6 Flash | high | 1335 ±26 |
| 21 | Opus 4.8 | max | 1334 ±20 |
| 22 | Muse Spark | 1334 ±28 | |
| 23 | Opus 4.8 | high | 1331 ±20 |
| 24 | Grok 4.5 | max | 1330 ±23 |
| 25 | GPT-5.5 | xhigh | 1329 ±17 |
| 26 | Kimi K2.6 | max | 1326 ±21 |
| 27 | GPT-5.5 | high | 1325 ±17 |
| 28 | GPT-5.4 | high | 1316 ±17 |
| 29 | Sonnet 5 | high | 1316 ±22 |
| 30 | Gemini 3 Flash | high | 1314 ±14 |