LMArena Vision: recognizing things leaderboard
- models measured
- 33
- lead over #2
- +22 Elo
- median
- 1226
Leader on LMArena Vision: recognizing things · read October 22
On October 22, 2026, 33 of our catalog's models had a LMArena Vision: recognizing things result, and Google Gemini 3.1 Pro Preview led with 1330, ahead of Google Gemini 3.1 Flash Lite Preview at 1308.
Image prompts asking who or what is shown. People ask a question, compare two anonymous answers and vote for the better one; the rating is built from those votes, as LMArena publishes it.
Role in the index
LMArena Vision: recognizing things is shown as published and does not feed any of our indexes. Read the methodology
| # | Model | Result | Votes | Interval (±) | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|---|---|
| 1 | Gemini 3.1 Pro PreviewGoogle | 1330 | 388 | ±28.9 | 32.3 | $12.00 | |
| 2 | Gemini 3.1 Flash Lite PreviewGoogle | 1308 | 259 | ±33.4 | 22.7 | $1.50 | |
| 3 | Kimi K2.5Moonshot AI | 1285 | 275 | ±32.1 | 38.6 | $2.85 | |
| 4 | Gemini 2.5 ProGoogle | 1283 | 917 | ±22.1 | 19.1 | $10.00 | |
| 5 | Claude Opus 4.7Anthropic | 1278 | 154 | ±40.6 | 67.5 | $25.00 | |
| 6 | Claude Opus 4.6Anthropic | 1274 | 189 | ±38.4 | 57.2 | $25.00 | |
| 7 | Qwen3.5 397B A17BQwen | 1265 | 222 | ±36.7 | Not benchmarked | $2.34 | |
| 8 | GPT-5OpenAI | 1263 | 432 | ±30.1 | 38.8 | $10.00 | |
| 9 | Gemini 2.5 FlashGoogle | 1252 | 628 | ±23.8 | Not benchmarked | $2.50 | |
| 10 | GPT-5.4OpenAI | 1251 | 137 | ±45.9 | 64.4 | $15.00 | |
| 11 | Qwen3 VL 235B A22B InstructQwen | 1249 | 151 | ±41.7 | Not benchmarked | $1.90 | |
| 12 | GLM 5V TurboZ.ai | 1241 | 231 | ±38.1 | Not benchmarked | $4.00 | |
| 13 | o3OpenAI | 1234 | 494 | ±29.8 | Not benchmarked | $8.00 | |
| 14 | GPT-5.2OpenAI | 1232 | 168 | ±41.1 | 55.9 | $14.00 | |
| 15 | GPT-5.1OpenAI | 1232 | 112 | ±49.4 | 40.4 | $10.00 | |
| 16 | GPT-5.5OpenAI | 1231 | 119 | ±52.3 | 71.3 | $30.00 | |
| 17 | GPT-5 MiniOpenAI | 1226 | 282 | ±35.6 | 34.2 | $2.00 | |
| 18 | Gemma 4 31BGoogle | 1226 | 359 | ±28.7 | 39.0 | $0.40 | |
| 19 | GPT-5.4 MiniOpenAI | 1218 | 183 | ±40.5 | 35.8 | $4.50 | |
| 20 | o4 MiniOpenAI | 1218 | 388 | ±32.0 | Not benchmarked | $4.40 | |
| 21 | Grok 4.3xAI | 1217 | 142 | ±44.7 | 37.4 | $2.50 | |
| 22 | Qwen3.5-27BQwen | 1210 | 187 | ±41.1 | Not benchmarked | $1.56 | |
| 23 | GPT-4.1OpenAI | 1208 | 392 | ±32.2 | Not benchmarked | $8.00 | |
| 24 | MiniMax M3MiniMax | 1197 | 107 | ±53.1 | 39.2 | $1.20 | |
| 25 | Gemma 3 27BGoogle | 1196 | 313 | ±34.3 | Not benchmarked | $0.30 | |
| 26 | Claude Sonnet 4.6Anthropic | 1195 | 162 | ±45.6 | 57.3 | $15.00 | |
| 27 | Gemma 4 26B A4BGoogle | 1191 | 235 | ±36.1 | Not benchmarked | $0.60 | |
| 28 | MiMo-V2.5Xiaomi | 1189 | 185 | ±40.3 | 50.0 | $0.336 | |
| 29 | GPT-4.1 MiniOpenAI | 1187 | 357 | ±32.8 | Not benchmarked | $1.60 | |
| 30 | Mistral Small 3.1 24BMistral AI | 1155 | 240 | ±40.8 | Not benchmarked | $0.555 | |
| 31 | Mistral Medium 3.1Mistral AI | 1147 | 439 | ±31.1 | Not benchmarked | $2.00 | |
| 32 | GPT-5.4 NanoOpenAI | 1141 | 178 | ±42.4 | 44.8 | $1.25 | |
| 33 | Mistral Medium 3Mistral AI | 1141 | 200 | ±40.2 | Not benchmarked | $2.00 |
Only models in our catalog are listed, and each appears once with its best published run. 184 rows were read from the source on October 22, 2026, and rows that match no catalog model are left out. Every other column is a figure the board publishes for the same run.