Name: 视觉理解模型排行榜（图像描述模型2026-07-02排行榜）最好的图像与多模态模型 |AITNT
Creator: LM Arena
License: https://creativecommons.org/licenses/by/4.0/
Keywords: 图像描述,AI 视觉理解模型排行榜,视觉 AI 排行榜,最佳图像与多模态模型,AI 模型排行

视觉模型 Arena 榜单LM Arena Global AI Leaderboard图像描述🖼️Captioning

比较并探索基于图像描述表现进行 Elo 积分排名的主流视觉 AI 模型，数据来源于真实用户匿名盲测。Explore Elo rankings of multimodal AI models that reason about images, charts, and OCR inputs — rated by real anonymous user votes.

2026-07-02

▾

上榜模型Models

参赛厂商Providers

8.2K

总投票数Total Votes

📖 指标说明 — 读懂这张榜单📖 Metrics Guide — Understanding the Leaderboard

🎯

榜单主指标（Score）Primary Metric (Score)

每个榜单会采用官网对应的核心排序指标。通用 Arena 榜单通常是模型竞技积分；智能体榜单则展示净提升及多项任务指标。

Each leaderboard uses the official primary ranking metric. Most Arena boards use competitive scores; the Agent board uses net improvement and task-specific metrics.

🔢

精确值（Float Value）Float Value

主指标的完整浮点数值。对于通用 Arena 榜单可用于区分整数分相近的模型；智能体榜单以百分比和置信区间展示。

The full floating-point value of the primary metric. It helps distinguish close scores on standard boards; Agent metrics are shown as percentages with confidence intervals.

📊

方差（Variance）Variance

衡量结果的统计稳定程度。绿色·稳定（<5）更可信；橙色·波动（5~12）；红色·不稳（>12）说明排名还可能明显变化。

Measures statistical stability. Green·Stable (<5); Orange·Unstable (5–12); Red·Volatile (>12) means the ranking may still move significantly.

↔️

置信区间（95% CI = ±N）95% Confidence Interval (CI = ±N)

95% 置信区间反映当前估计的不确定范围，显示为 ±N。在相同计算口径下，N 越小说明估计越集中、排名越稳定。

The 95% confidence interval, shown as ±N, summarizes uncertainty in the current estimate. Under the same methodology, a smaller N indicates a more precise estimate and a more stable ranking.

🗳️

投票数（Vote Count）Vote Count

该模型或厂商参与评测的总次数。数量越高，结果的统计置信度越高、越不容易受单次样本影响而波动。

Total evaluation volume for the model or lab. Higher counts mean higher statistical confidence and less sensitivity to individual samples.

📜

开源协议（License）License

Apache/Llama：完全开源可商用；MIT：开源限制极少；Proprietary：闭源商业模型。协议决定你能否把它集成到自己的产品里。

Apache/Llama: Fully open-source, commercial use allowed; MIT: Open-source, minimal restrictions; Proprietary: Closed-source commercial model. License determines if you can integrate it into your product.

排名Rank	名次区间Rank Range	模型Model		投票数Votes	精确分 & 区间Float Score & CI	方差Variance	价格 ($/1M)Price ($/1M)	上下文Context
🥇	1 - 18	gemini-3.1-pro-preview Google · Proprietary	1274±51.0	116 票votes	1274.000 [1223.0, 1325.0]	17.00 · 不稳Volatile	$2 / $12	1M
🥈	1 - 14	gemini-3-pro Google · Proprietary	1273±36.0	222 票votes	1273.000 [1237.0, 1309.0]	13.00 · 不稳Volatile	$2 / $12	1M
🥉	1 - 20	gpt-5.2-high OpenAI · Proprietary	1268±60.0	83 票votes	1268.000 [1208.0, 1328.0]	19.00 · 不稳Volatile	$1.75 / $14	400K
4	1 - 17	gemini-2.5-pro Google · Proprietary	1251±23.0	809 票votes	1251.000 [1228.0, 1274.0]	16.00 · 不稳Volatile	$1.25 / $10	1M
5	1 - 25	gpt-5.1-high OpenAI · Proprietary	1244±56.0	99 票votes	1244.000 [1188.0, 1300.0]	24.00 · 不稳Volatile	$1.25 / $10	400K
6	1 - 25	gemini-3.1-flash-lite-preview Google · Proprietary	1242±72.0	57 票votes	1242.000 [1170.0, 1314.0]	24.00 · 不稳Volatile	$0.25 / $1.50	1M
7	1 - 25	gemini-3-flash Google · Proprietary	1230±41.0	173 票votes	1230.000 [1189.0, 1271.0]	24.00 · 不稳Volatile	$0.50 / $3	1M
8	1 - 25	chatgpt-4o-latest-20250326 OpenAI · Proprietary	1214±35.0	284 票votes	1214.000 [1179.0, 1249.0]	24.00 · 不稳Volatile	$5 / $15	128K
9	1 - 25	gemini-2.5-flash Google · Proprietary	1214±25.0	596 票votes	1214.000 [1189.0, 1239.0]	24.00 · 不稳Volatile	$0.30 / $2.50	1M
10	1 - 25	gpt-4.1-2025-04-14 OpenAI · Proprietary	1209±30.0	440 票votes	1209.000 [1179.0, 1239.0]	24.00 · 不稳Volatile	$2 / $8	1M
11	1 - 27	kimi-k2.5-thinking 月之暗面 · Modified MIT	1207±54.0	103 票votes	1207.000 [1153.0, 1261.0]	26.00 · 不稳Volatile	$0.60 / $3	N/A
12	1 - 26	qwen3-vl-235b-a22b-instruct 阿里巴巴 · Apache 2.0	1207±46.0	149 票votes	1207.000 [1161.0, 1253.0]	25.00 · 不稳Volatile	$0.20 / $0.88	262.1K
13	1 - 27	gemini-3-flash (thinking-minimal) Google · Proprietary	1206±46.0	151 票votes	1206.000 [1160.0, 1252.0]	26.00 · 不稳Volatile	$0.50 / $3	1M
14	2 - 25	gpt-5-chat OpenAI · Proprietary	1199±30.0	398 票votes	1199.000 [1169.0, 1229.0]	23.00 · 不稳Volatile	$1.25 / $10	128K
15	2 - 26	gpt-4.1-mini-2025-04-14 OpenAI · Proprietary	1197±32.0	407 票votes	1197.000 [1165.0, 1229.0]	24.00 · 不稳Volatile	$0.40 / $1.60	1M
16	3 - 27	gpt-5-high OpenAI · Proprietary	1190±34.0	379 票votes	1190.000 [1156.0, 1224.0]	24.00 · 不稳Volatile	$1.25 / $10	400K
17	4 - 27	gemini-2.5-flash-lite-preview-06-17-thinking Google · Proprietary	1189±31.0	401 票votes	1189.000 [1158.0, 1220.0]	23.00 · 不稳Volatile	$0.10 / $0.40	1M
18	2 - 29	gpt-5.1 OpenAI · Proprietary	1185±47.0	125 票votes	1185.000 [1138.0, 1232.0]	27.00 · 不稳Volatile	$1.25 / $10	400K
19	1 - 29	gemma-4-31b Google · Apache 2.0	1184±69.0	64 票votes	1184.000 [1115.0, 1253.0]	28.00 · 不稳Volatile	$0.14 / $0.40	262.1K
20	5 - 27	o3-2025-04-16 OpenAI · Proprietary	1179±28.0	555 票votes	1179.000 [1151.0, 1207.0]	22.00 · 不稳Volatile	$2 / $8	200K
21	5 - 28	o4-mini-2025-04-16 OpenAI · Proprietary	1178±30.0	439 票votes	1178.000 [1148.0, 1208.0]	23.00 · 不稳Volatile	$1.10 / $4.40	200K
22	5 - 29	gpt-5-mini-high OpenAI · Proprietary	1165±41.0	302 票votes	1165.000 [1124.0, 1206.0]	24.00 · 不稳Volatile	$0.25 / $2	400K
23	5 - 29	grok-4-0709 xAI · Proprietary	1164±35.0	374 票votes	1164.000 [1129.0, 1199.0]	24.00 · 不稳Volatile	$3 / $15	256K
24	4 - 30	gpt-5.2 OpenAI · Proprietary	1156±60.0	92 票votes	1156.000 [1096.0, 1216.0]	26.00 · 不稳Volatile	$1.75 / $14	400K
25	5 - 30	gemini-2.0-flash-001 Google · Proprietary	1146±58.0	110 票votes	1146.000 [1088.0, 1204.0]	25.00 · 不稳Volatile	$0.10 / $0.40	1M
26	14 - 29	mistral-medium-2508 Mistral · Proprietary	1131±29.0	411 票votes	1131.000 [1102.0, 1160.0]	15.00 · 不稳Volatile	$0.40 / $2	131.1K
27	12 - 30	mistral-small-3.1-24b-instruct-2503 Mistral · Apache 2.0	1129±37.0	278 票votes	1129.000 [1092.0, 1166.0]	18.00 · 不稳Volatile	$0.10 / $0.30	32K
28	19 - 30	gemma-3-27b-it Google · Gemma	1112±38.0	270 票votes	1112.000 [1074.0, 1150.0]	11.00 · 波动Unstable	$0.08 / $0.16	131.1K
29	20 - 30	mistral-medium-2505 Mistral · Proprietary	1098±48.0	166 票votes	1098.000 [1050.0, 1146.0]	10.00 · 波动Unstable	$0.40 / $2	131.1K
30	25 - 30	mistral-small-2506 Mistral · Apache 2.0	1052±47.0	189 票votes	1052.000 [1005.0, 1099.0]	5.00 · 波动Unstable	$0.10 / $0.30	32K

🔍

没有找到匹配的模型，换个关键词试试？No matching models found. Try a different keyword?