Google rebranded its public AI chatbot formerly known as "Bard" to "Gemini" last week. But reviews so far are mixed.

Google now has its second answer to OpenAI’s ChatGPT, the internet’s preeminent AI chatbot. Users can try out a free version of Gemini, a combination of search engine, voice assistant, and image generator, or pay for Gemini Advanced, which also has productivity tools that integrate into the Google suite and a more advanced back-end model.

"With our Ultra 1.0 model, Gemini Advanced is far more capable at highly complex tasks like coding, logical reasoning, following nuanced instructions and collaborating on creative projects," Googl says in a blog post. "Gemini Advanced not only allows you to have longer, more detailed conversations; it also better understands the context from your previous prompts."

As TechCrunch reports, Gemini Advanced’s latest models have some impressive features. It can take in ~700,000 words, ~30,000 lines of code, and up to 11 hours of audio or an hour of video in several different languages.

But tech reporters who tested out Gemini weren’t overly impressed.

TechCrunch’s Kyle Wiggers asked it for the lastest updates in the Israel-Palestine conflict, and got an answer that recommended he Google it instead. It also sidestepped questions about contentious topics like immigration and college admissions. Gemini even told some pretty humorless jokes. Still, Wiggers liked the plugin to his Gmail inbox.  

The Verge’s Emilia David similarly tested Gemini’s features. She asked it to generate an image of a dog, and got one with two tongues and an extra limb (a classic AI art issue).

SFist asked Gemini to draw a picture of "modern San Francisco" and this was one of the results, below. What bridge is that?

Gemini's version of "modern San Francisco."

The chatbot also hallucinated when it tried to give information like addresses of local restaurants, David says, and claimed to known the Filipino language but would not translate a phrase in Filipino — and Filipino is not listed as a language Gemini knows. But David was impressed with its ability to synthesize complex research topics.

"Gemini Advanced is capable. There’s no denying it works much better than the lower-tier Gemini," David writes. "It definitely works best when integrated with Google’s other products like Search and Maps. But for more obviously 'creative' multimodal requests — things that involve images, for example — Gemini has a long way to go."

As Wiggers writes, “Paying $20 a month for Ultra right now feels like a big ask.”

Plus, there are still privacy concerns about the data that users input as Google continues to improve its models. It’s early days, but this is another case where it doesn’t feel like AI is about to take over the world just yet.

Related: Google Launches Updated AI Chatbot That They Say Is More Powerful Than ChatGPT