Google Upgrades Gemini 2.5 Flash With Improved Image Understanding and Better Formatting

Google is improving the Gemini 2.5 Flash artificial intelligence (AI) model. On Thursday, the Mountain View-based tech giant improved several capabilities of the model, making its responses more organised and making it more capable at analysing and answering questions about uploaded images. Additionally, the large language model (LLM) will also be able to better help students with their homework. All of these capabilities are currently rolling out to users globally. It will be available across both mobile apps as well as the website.

Gemini 2.5 Flash Is Now Better At Presenting Answers to Queries

In a series of posts on X, the official handle of the Gemini app listed the improvement the company is making to the Gemini 2.5 Flash model. Notably, it is the default AI model for those on the free tier and is available to both Google AI Pro and Ultra subscribers. Unlike the Pro version, it is designed to provide faster responses while maintaining a high level of reasoning.

The improvements have been made in three focus areas. First is the improvement in homework assistance. Google says the Gemini 2.5 Flash's reasoning capabilities have been improved, enabling it to respond to complex homework questions. Following the Socratic method of learning, the chatbot will guide students in a step-by-step manner, breaking down complex topics into easy-to-understand portions.

Gemini is also getting better at generating scannable outputs. Scannable outputs essentially mean the responses that are formatted in a way that makes it easy to skim through the information. The AI model will now use elements such as headers, lists, and tables to make the responses more digestible.

Finally, the tech giant has also improved the AI model's image understanding. Image understanding is part of the computer vision functionality, and enables an LLM to process different elements in a photo and contextually understand them. With the new upgrade, the company says the model can now understand and answer queries about a detailed image or diagram.

Apart from this, Google has also added a new quality-of-life feature. Users can now upload their rough notes and ask the AI model to organise or summarise the information. Additionally, the Gemini 2.5 Flash model can create flashcards for users.