OpenAI o3/o4-mini OpenAI Image thinking chain 2025.4
The first model with image thinking chain reasoning capabilities, a revolutionary breakthrough in vision and programming
🧠 What are o3 and o4-mini?
o3 and o4-mini is the latest inference model released by OpenAI on April 17, 2025. This is the first in the o series to feature Image thinking chain reasoning Products with powerful capabilities mark AI’s breakthrough progress in the field of visual understanding.
👁️ Image thinking chain
Look at images and perform in-depth reasoning, realizing the visual thinking chain for the first time
💻 Coding ability
Mathematics, programming, and scientific exploration surpass the previous generation of o1
🔧 Tool usage
Native support for tool calls to build complex Agent applications
⚡ Codex CLI
Official terminal tool, released less than 7 hours ago, Star exceeds 5500
📊 Model series comparison
o3 Flagship Reasoning
The most powerful inference model, significantly surpassing the previous generation o1 in many fields such as mathematics, programming, scientific exploration, and visual recognition. Supports complete image thinking chain reasoning.
o4-mini Efficient reasoning
A version optimized for fast and efficient inference, maintaining strong capabilities while significantly reducing latency and cost. Free users can submit queries to try.
o3-pro Coming soon
An enhanced version planned to be released in the coming weeks that supports the full functionality of the tool and provides more powerful capabilities for Pro users.
👁️ Image thinking chain reasoning
Image thinking chain is the most important innovation of o3/o4-mini. The traditional visual model can only describe what is seen, while the image thinking chain can:
📐 Math Charts
Analyze geometric figures, function images, and perform mathematical reasoning
📝 Handwriting recognition
Quickly recognize handwritten text and automatically adjust pictures to obtain information
💻 Code screenshot
Understand code screenshots, analyze bugs, and provide repair suggestions
📄 Document analysis
Parse complex documents, tables, flow charts, etc.
💻 Codex CLI - Terminal Programming Assistant
Codex CLI, released simultaneously with o3/o4-mini, is the official terminal programming tool of OpenAI. The number of GitHub stars has exceeded 5,500 in less than 7 hours since its release.
# 安装 Codex CLI npm install -g @openai/codex-cli # 使用 o3 模型进行代码生成 codex "用 Python 实现一个支持并发的网络爬虫" # 分析代码截图 codex --image ./screenshot.png "解释这段代码的问题" # 交互式会话 codex chat --model o3
🖥️ Terminal native
Interact with o3/o4-mini directly in the terminal
🖼️ Image support
Support uploading code screenshots for analysis
📁 File operations
Directly read and write files and execute commands
🎯 Typical application scenarios
📊 Data visualization analysis
Upload chart screenshots, AI automatically analyzes data trends and insights
🐛 Debug Assistant
Upload error screenshots, quickly locate the problem and provide a repair plan
📱 UI/UX Analysis
Analyze interface screenshots and provide design improvement suggestions
🔬 Scientific research
Analyze experimental charts and paper images to assist research work
🔗 How to access
| User type | o3 | o4-mini | o3-pro |
|---|---|---|---|
| Free user | ❌ | ✅ LIMITED LIMITED | ❌ |
| Plus User | ✅ | ✅ | ❌ |
| Pro User | ✅ | ✅ | ✅ Coming soon |
| API Developer | ✅ | ✅ | Coming soon |