Elon Musk's AI start-up xAI has released an improved version of Grok. According to internal benchmarks, the current Grok 2 model performs better than GPT-4o in some tests. Only Anthropic's Claude with the Sonnet 3.5 model is ahead of it.
Grok 2 is now available to all X users. Until then, you had to have a paid premium subscription to use the chatbot.
Real-time information and image generation
In parallel with wider availability, there are also more functions. This means Grok 2 can access information in real time, the source for this being both Platform X and the web. Sources are cited so that users can understand the origin of the information.
Grok 2 quotes extensively from sources. How decoder reported, but it is not yet known whether xAI has entered into partnerships with media and press publishers. Competing providers such as OpenAI now have numerous such agreements.
It is now also possible to generate images directly with Grok. And in the future, a Grok switch will be integrated into X posts, which can then be used to analyze the content of a post. This should provide users with relevant context for the respective post as well as ongoing discussions and trends.
Grok-2 model is expected to outperform GPT-4o
xAI now offers access to the current Grok 2 and Grok 2 Vision models via the API. According to an internal benchmark, progress was once again achieved compared to the beta versions. Grok 2 takes second place and is ahead of GPT-4o and the preview version of o1 – the final o1 version, which OpenAI recently presented, has not yet been tested. Meanwhile, Anthropic takes first place with the Claude 3.5 Sonnet model.
xAI was also able to improve efficiency. This means that the price per token drops when users access the models via the API.
