> For the complete documentation index, see [llms.txt](https://docs.cherryai.com.cn/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.cherryai.com.cn/docs/zhong-wen-fan-ti/other/model_rank/vision.md).
# 視覺榜單
本頁展示 [Arena AI](https://arena.ai/leaderboard/vision) 榜單前 30 名嘅每日快照，方便快速了解近期模型表現。完整榜單、篩選項同最新變化請到 Arena AI 官網查看。
本榜單評測多模態模型喺圖像理解類任務上嘅表現。
> **數據更新時間**：2026-09-23 13:18:19 UTC / 2026-09-23 21:18:19 CST (北京時間)
{% hint style="info" %}
排行榜反映特定評測同用戶投票偏好，唔等同於模型喺你嘅任務入面一定更好。選擇模型時仲要考慮價格、速度、上下文、工具調用、私隱同地區可用性。
{% endhint %}
## 前 30 名
| 排名 | 排名區間 | 模型 | 分數 | 票數 | 價格 $/百萬Token | 上下文 |
| -: | ---- | -------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------- | -----: | ------------- | ---- |
| 1 | 1-8 | claude-fable-5-high [1](https://www.anthropic.com/news/claude-fable-5-mythos-5) | 1310 (±8) | 11,304 | $10 / $50 | 1M |
| 2 | 1-14 | qwen3.8-max [1](https://www.qwencloud.com/models/qwen3.8-max) | 1302 (±8) | 8,665 | $1.69 / $5.07 | 1M |
| 3 | 1-14 | claude-opus-4-7-high [1](https://www.anthropic.com/news/claude-opus-4-7) | 1301 (±7) | 21,092 | $5 / $25 | 1M |
| 4 | 1-15 | claude-opus-4-7 [1](https://www.anthropic.com/news/claude-opus-4-7) | 1300 (±7) | 21,450 | $5 / $25 | 1M |
| 5 | 1-16 | claude-opus-4-6-high [1](https://www.anthropic.com/news/claude-opus-4-6) | 1299 (±7) | 20,835 | $5 / $25 | 1M |
| 6 | 1-30 | muse-spark-1.3-max [1](https://developer.meta.com/ai/models/muse-spark/) | 1294 (±15) | 1,804 | N/A | N/A |
| 7 | 1-28 | muse-spark [1](https://ai.meta.com/blog/introducing-muse-spark-msl/) | 1294 (±9) | 5,572 | N/A | N/A |
| 8 | 2-26 | claude-opus-4-6 [1](https://www.anthropic.com/news/claude-opus-4-6) | 1293 (±7) | 25,140 | $5 / $25 | 1M |
| 9 | 1-31 | muse-spark-1.2 (xHigh) [1](https://developer.meta.com/ai/models/muse-spark/) | 1292 (±15) | 1,841 | $1.25 / $4.25 | N/A |
| 10 | 2-30 | claude-opus-5-high [1](https://www.anthropic.com/news/claude-opus-5) | 1289 (±8) | 11,599 | $5 / $25 | 1M |
| 11 | 2-31 | claude-fable-5.1-max [1](https://www.anthropic.com/claude-fable-and-mythos-5-1) | 1289 (±12) | 2,701 | $10 / $50 | 1M |
| 12 | 2-30 | gemini-3-pro [1](http://aistudio.google.com/app/prompts/new_chat?model=gemini-3-pro-preview) | 1289 (±8) | 12,995 | $2 / $12 | 1M |
| 13 | 4-30 | gpt-5.5 [1](https://openai.com/index/introducing-gpt-5-5/) | 1287 (±6) | 23,423 | $5 / $30 | 1.1M |
| 14 | 2-31 | gpt-5.6-sol-xhigh [1](https://openai.com/index/gpt-5-6/) | 1286 (±8) | 7,729 | $4 / $20 | N/A |
| 15 | 6-30 | gpt-5.4-high [1](https://platform.openai.com/docs/models/gpt-5.4) | 1285 (±6) | 25,357 | $2.50 / $15 | 1.1M |
| 16 | 2-38 | gpt-6-astra-max [1](https://openai.com/index/gpt-6-astra/) | 1284 (±17) | 1,367 | N/A | N/A |
| 17 | 6-31 | gemini-3.5-flash-high [1](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5/) | 1284 (±8) | 9,579 | $0.75 / $4.50 | 1M |
| 18 | 6-31 | claude-opus-4-8-high [1](https://www.anthropic.com/news/claude-opus-4-8) | 1283 (±7) | 15,274 | $5 / $25 | 1M |
| 19 | 6-31 | gpt-5.5-high [1](https://openai.com/index/introducing-gpt-5-5/) | 1283 (±6) | 21,960 | $5 / $30 | 1.1M |
| 20 | 6-32 | gemini-3.5-flash-medium [1](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5/) | 1283 (±8) | 9,618 | $0.75 / $4.50 | 1M |
| 21 | 5-35 | gemini-3.6-flash-high [1](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-6-flash-3-5-flash-lite-3-5-flash-cyber/) | 1283 (±10) | 5,159 | $0.75 / $3.75 | 1M |
| 22 | 6-35 | muse-spark-1.1 [1](https://ai.meta.com/blog/introducing-muse-spark-meta-model-api/) | 1281 (±8) | 8,447 | $1.25 / $4.25 | 1M |
| 23 | 6-35 | gpt-5.4 [1](https://platform.openai.com/docs/models/gpt-5.4) | 1280 (±7) | 21,188 | $1.25 / $7.50 | 1.1M |
| 24 | 6-35 | grok-4.5 [1](https://docs.x.ai/developers/models/grok-4.5) | 1279 (±8) | 8,355 | $2 / $6 | 500K |
| 25 | 7-35 | claude-opus-4-8 [1](https://www.anthropic.com/news/claude-opus-4-8) | 1279 (±7) | 15,801 | $5 / $25 | 1M |
| 26 | 8-35 | gemini-3.1-pro-preview [1](https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/) | 1279 (±5) | 40,691 | $1 / $6 | 1M |
| 27 | 7-35 | gpt-5.2-chat-latest-20260210 [1](https://developers.openai.com/api/docs/models/gpt-5.2-chat-latest) | 1278 (±7) | 15,432 | $1.75 / $14 | 128K |
| 28 | 6-37 | gpt-5.5-instant [1](https://openai.com/index/gpt-5-5-instant) | 1278 (±9) | 7,226 | $5 / $30 | 1.1M |
| 29 | 8-37 | claude-sonnet-4-6 [1](https://www.anthropic.com/news/claude-sonnet-4-6) | 1275 (±6) | 25,552 | $1.50 / $7.50 | 1M |
| 30 | 6-41 | glm-5.3-flash [1](https://z.ai/blog/glm-5.3) | 1275 (±12) | 3,110 | $0.06 / $0.20 | 1M |
## 呢張表點睇
* **排名 / 排名區間**：Arena AI 根據對戰投票估算嘅相對名次；排名區間表示喺置信範圍內嘅名次波動。
* **分數**：相對評分，適合比較同一時刻榜單入面嘅模型。
* **票數 / 會話數**：樣本量參考；樣本較少時，排名通常更容易波動。
* **價格 $/百萬Token**：輸入 / 輸出每百萬 Token 嘅參考價格。
* **上下文**：模型支援嘅最大上下文長度。
## 揀模型嗰陣再確認三件事
1. Provider 係咪真係提供呢個模型，同埋你嘅地區同帳戶係咪可用；
2. API 價格、速率限制同上下文係咪適合你嘅任務；
3. 用 3～5 個真實任務做小規模測試，唔好淨係睇總榜名次。
## 數據來源
數據來自 [Arena AI 官方 視覺榜單](https://arena.ai/leaderboard/vision)，由 GitHub Actions 每日更新。模型價格、許可證同能力請以模型服務商官方資訊為準。
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.
Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:
```
GET https://docs.cherryai.com.cn/docs/zhong-wen-fan-ti/other/model_rank/vision.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.