Google’s free Gemini chatbot gets 1.5 Flash update, making responses faster and smarter

Google LLC said today that its Gemini AI chatbot is being updated with support for Gemini 1.5 Flash, a light-weight large language model that’s designed to compete with OpenAI’s GPT-4o mini.

Gemini 1.5 Flash is being made accessible to all users, no matter their subscription level. With the update, Google said, users will see “across-the-board” improvements almost about the standard of Gemini AI’s response quality and the speed at which it delivers those responses. They’ll also notice improvements in its reasoning and image understanding capabilities, it promised.

Google first showcased Gemini 1.5 Flash during its Google I/O 2024 event in May and made it available to some paying customers shortly after. It’s now landing within the free version of the Gemini mobile app, and likewise the freely accessible web interface at gemini.google.com.

The improved capabilities of Gemini 1.5 Flash stem from a large boost in token size. It now supports 32,000 tokens, 4 times greater than the unique version of the synthetic intelligence model that debuted in May. The increased variety of tokens in Gemini 1.5 Flash means it might probably handle longer, more complex prompts and deliver higher responses, the corporate said.

“Meaning you possibly can have longer back-and-forth conversations and ask Gemini more complex questions — all freed from charge,” Amar Subramanya, vice chairman of engineering for Gemini Experiences, said in a blog post.

As well as, Subramanya said, users will soon have the ability to upload files to the free version of its Gemini AI chatbot, which was previously possible only within the paid version. Uploading files is completed to support contextual prompts. For example, it’s possible to upload a picture and ask the model questions on it, or feed it an economics study guide and ask the model to create practice questions.

One other soon-to-launch feature will enable Gemini to investigate data files and create visualizations and charts based on them.

Subramanya said Google can also be taking steps to scale back AI hallucinations, which is the term used to explain nonsensical or inaccurate responses. To do that, Gemini will provide citations to all of its responses, linking to all the source materials it used. So if the model comes up with a questionable response, users will have the ability to explore that source and take a look at to ascertain the validity of its claims by themselves. Subramanya said these citations will even extend to those using the Gmail extension, which makes it possible to ask Gemini questions pertaining to their email inboxes.

The Gemini chatbot can also be being added to Google Messages within the European Economic Area, the U.K. and Switzerland. Users will have the ability to pick “Start chat” throughout the Messages app and choose “Gemini” and immediately start interacting with the chatbot.

Finally, Google said it should allow teenagers as young as 13 to begin using Gemini. Based on Subramanya, teens who’ve a Google Account can safely use Gemini as a research tool, as the corporate has worked with child safety organizations to implement latest policies and safeguards against misuse.

Image: Google