"Claude" please sit at the kids' table! Gemini 2.5 Pro is released, the best coding tool!

Google recently released Gemini 2.5 Pro (I/O version), which is currently the most powerful AI model in coding and leads in many programming tests.

Encoding performance exceeds all competitors

Google said that Gemini 2.5 Pro ranked first on the WebDev Arena leaderboard. This ranking specifically tests the model's ability to build beautiful and fully functional web applications. Simply put, this model has surpassed all competitors in writing front-end code, including industry-leading models such as Claude 3.7, Claude 3.5, GPT-4.1, and DeepSeek-V3/R1 .

"We found the Gemini 2.5 Pro to be the best leading-edge model in terms of 'power to latency'," said Michele Catasta, president of Replit.

Silas Alberti from the Cognition founding team also commented: "The upgraded Gemini 2.5 Pro performed very well in our junior developer evaluation. It was the first model to solve one of our evaluations involving a large backend request routing refactor. It acts like a more experienced developer, making good judgments and choosing good abstractions. "

One netizen joked: "Claude can sit at the children's table now!"

According to the top 10 most popular AI models released by Cursor, Gemini 2.5 Pro ranked second and will now make a sprint towards the first place!

Coding new features to make development easier

Gemini 2.5 Pro brings several key coding enhancements:

1. Video to code conversion

The new version scored 84.8% on the VideoMME benchmark, showing strong video comprehension capabilities. Combined with encoding capabilities, it can create a complete interactive learning application based on a YouTube video.

2. Rapid feature development

Front-end development usually requires manually reviewing design files, checking components to match style properties, and then writing CSS code. Now, Gemini 2.5 Pro can help you generate new features in the IDE, such as adding a video player according to the existing style of the application.

Performance is overwhelming

According to the data released by Google, the Gemini 2.5 Pro achieved impressive scores in almost all benchmarks:

• WebDev Arena (Web Development): Ranked #1
• Global MMLU (Lite): 88.6%
• VideoMME (video understanding): 84.8%
• Complex reasoning ability: far beyond competitors

Get started now

Developers can start using Gemini 2.5 Pro through the Gemini API in Google AI Studio , and enterprise customers can use it through Vertex AI. For developers who are already using Gemini 2.5 Pro, this new version will not only improve coding performance, but also solve some problems in previous versions, including reducing function call errors and improving function call trigger rates.

access https://aistudio.google.com/prompts/new_chat As shown in the following interface:

According to the latest official information, the pricing of Gemini 2.5 Pro Preview 05-06 is as follows:

• Enter price :

• 200K tokens or less: $1.25/1M tokens
• More than 200K tokens: $2.50/1M tokens

• Output price :

• Within 200K tokens: $10.00/1M tokens
• More than 200K tokens: $15.00/1M tokens

• Daily free quota : 5 RPM (requests per minute) per day, up to 25 requests per day

• Rate limit : 150 RPM (requests per minute)

The API price is calculated per 1M tokens, and the UI interface remains free to use.

Actual Combat - Space Games

To demonstrate the coding capabilities of Gemini 2.5 Pro, let's take a look at a real-world example: creating a space shooting game.

I made the following requirements to Gemini 2.5 Pro:

Create a space shooting game with a spaceship that can be controlled with the arrow keys and can shoot aliens. The player's spaceship and alien enemies can be drawn with simple patterns to create a dynamic starry sky background to increase the feeling of space. When there are aliens in the universe, add a simple particle explosion effect

With just a simple prompt, Gemini 2.5 Pro can generate a complete, runnable game code.

This code creates a complete space shooter game, including:

1. A triangular player ship that can be controlled with the arrow keys
2. Press the space bar to fire bullets
3. Square alien enemy with eyes and mouth
4. Dynamic starry sky background
5. Particle explosion effects when hitting aliens
6. Complete collision detection and game loop

The game effect is shown in the figure below:

The effect is much better than before. See Gemini2.5 Pro Canvas is free for everyone, and it’s really cool in actual testing!

In this way, even novice programmers can quickly create feature-rich applications, while professional developers can use the code as a basis to further optimize and expand the functionality.

Why is this upgrade so important?

AI coding assistants are changing the way software is developed. This upgrade of Gemini 2.5 Pro takes AI-assisted programming to a new level:

1. Better UI development : You can create more beautiful and functional front-end interfaces
2. Stronger reasoning ability : can understand complex code logic and system architecture
3. Multimodal input : supports text, image, video, and audio input to understand requirements more flexibly
4. Large context window : supports 1 million token inputs and can handle larger code bases

These capabilities not only improve development efficiency, but also lower the threshold for programming, allowing more people to realize their software ideas through AI.

Conclusion

The release of Gemini 2.5 Pro marks a new milestone in AI coding capabilities. Whether you are a professional developer or a programming novice, this tool can significantly improve your development efficiency and code quality.

In the future, we can expect AI coding assistants to play a role in more complex scenarios and further change the way software is developed. The future of programming is being reshaped by AI.