Google Gemini is learning to do a better job than Google Assistant. It acts more like a virtual assistant and is getting new updates to improve its conversational feature.
Google recently announced an update to Gemini 2.5 Flash Native Audio for live voice agents. The new update rolls out improvements in reliability, adhering to instructions and conversing smoothly.
Sharper function calling
Google Gemini is more reliable in triggering external functions. It is more accurate in identifying when to gather real-time information during a conversation. It can also include the data in its audio response seamlessly, keeping the conversation running smoothly.
Following instructions
Compared to the previous model that has an adherence rate of 84%, the updated model has a 90% chance of obeying developer instructions. This improvement allows it to handle complex instructions better and deliver more reliable outputs, instead of digressing from the topic of interest.
Smoother conversations
Gemini 2.5 Flash Native Audio can now retrieve context from previous points in the conversation more effectively, allowing for more cohesive conversations. You don’t have to repeat your previous messages as it can recall them and respond based on the stated points.
Other than these improvements, the VP of Google Labs, Gemini, and AI Studio, Josh Woodward, shared two other enhancements. One of them allows you to complete your statements without Gemini Live cutting you off mid-sentence if you pause for too long. The other lets you mute your microphone so you don’t interrupt Gemini Live while it is talking.
Google aims to make the model better at handling complex workflows, navigating user instructions, and holding smoother, flowing conversations. The update has started rolling out to Gemini Live, Search Live, Google AI Studio, and Vertex AI. The company also plans to bring a few changes to the Translate app.
