Grok 3 release: functional tests and GPT comparison

Today, X Platform released its major update, launching the Grok 3. As a user who is lucky enough to get early access, I will test this new AI model and compare it with ChatGPT. Grok 3 brings two new modes: deep search and thinking mode, which seem to embed some kind of inference model that can be easily switched between user interfaces.

First of all, the Grok 3's user interface is very similar to ChatGPT, and everything from brand colors to button icons makes people feel familiar. Despite this, we went straight to the testing phase.

Access and interface :

You can access the new version of Grok on grok.com, which is not only available on X platform, but also has a dedicated user interface.

Chats can be shared directly to X, making content dissemination more convenient.

Test comparison: Grok 3 performance reveals

In order to gain a deeper understanding of Grok 3, the video blogger's strength has conducted a series of comparison tests, covering multiple dimensions such as naming, coding, image generation, content creation, and online search, and combining Grok 3 with celebrities such as ChatGPT, DeepSeek, and Perplexity. Products compete on the same stage.

• Name: Grok 3 in the regular mode performed poorly in the "Dog Breed Classification ending in i" test and failed to accurately understand the instructions. However, after successfully switching to the inference mode, Grok 3 overcomes this challenge, gives a meeting of the requirements, and initially implements the advantages of the inference mode.

• Coding test: In the test of building the "Space Invaders and Snake Combination Game", although the inference mode of Grok 3 ended up thinking for quite a long time, the generated HTML code failed to run effectively, just create a green square. The encoding capability is slightly inferior to ChatGPT.

• Image Generation Test: In the "Generate Pictures of Giraffe Riding on the Magic Carpet", Grok 3 showed amazing image generation capabilities, with rich works and interesting details, even surpassing ChatGPT's performance, which was eye-catching.

• Content Creation (SEO Optimization Article) Test: In the test of "Create an SEO Optimization Article about SEO Training Japan", the articles generated by Grok 3 are no longer humane, the text is smooth and natural, the format and layout are also better, and the content length is slightly Compared with ChatGPT, it is better in content creation.

• Web Search/In-depth Research Test: Video bloggers conducted "In-depth Research" tests in Grok 3, Perplexity and ChatGPT, respectively. Grok 3's depth search speed is extremely fast, but the result is interrupted and the depth and data timeliness are not as good as Perplexity. Perplexity is equally ahead in depth and data sources, but the speed advantages of Grok 3 and Twitter's real-time data integration capabilities cannot be ignored.

Conclusion and evaluation :

Grok 3 performs well in speed and image generation, but still needs to be refined in terms of code execution and in-depth research.

For users who are subscribed to X Premium Services, Grok 3 offers some free query permissions, but premium features are available for an additional fee.

With the launch of Grok 3, it is not just a chatbot, but also attempts to become an all-round search and content creation platform.

Overall, although Grok 3 still needs to be improved in some ways, its innovation and potential are estimable, especially for users of the X platform, which is an exciting new tool.