Google rolls out Gemini 3.1
Google launched Gemini 3.1 Flash Live — a real‑time, multilingual voice AI with built‑in watermarking that's already topping the Hong Kong App Store and being embedded into commerce (Gap says it supports direct checkout inside Gemini). (eweek.com, firstpost.com, technobezz.com)
Gemini 3.1 Flash Live is being shipped in preview to developers through the Gemini Live API in Google AI Studio, offered to enterprises via Gemini Enterprise for Customer Experience, and is the audio engine behind Google’s conversational surfaces such as Search Live and Gemini Live. (blog.google, ai.google.dev) The Flash Live model is an audio‑to‑audio, streaming multimodal surface that ingests continuous audio, video and text and natively supports “more than 90 languages,” with Google citing a 90.8% score on ComplexFuncBench Audio and a 36.1% score on Scale AI’s Audio MultiChallenge. (blog.google, ai.google.dev) All outputs from the 3.1 Flash Live audio pipeline are watermarked with Google DeepMind’s SynthID system, and Google has added audio uploads to its SynthID Detector so users and moderators can check clips for the invisible watermark. (blog.google, deepmind.google, support.google.com) Gap Inc.’s recent announcement confirms its brands will appear “transaction‑ready” inside AI experiences by adopting Google’s Universal Commerce Protocol and acting as an early partner for Bold Metrics’ Agentic Sizing Protocol for embedded fit guidance. (gapinc.com, retailtouchpoints.com) Google says the Flash Live upgrade is enabling a global roll‑out of Search Live to “more than 200 countries and territories,” and the company has extended related real‑time interpreter/headphone modes to iOS in addition to Android. (blog.google, techcrunch.com, droid-life.com) System‑design interview prompt tied to this launch: sketch a low‑latency, streaming speech‑to‑speech architecture that integrates the Gemini Live API’s audio streaming surface, handles session/ephemeral tokens and function‑calling for tool access, and routes traffic across regions for global availability as documented in the Live API and Vertex AI docs. (ai.google.dev, cloud.google.com) PM case study grounded in the Gap integration: design a launch plan for in‑agent commerce using UCP that defines experiment KPIs (checkout conversion, average order value, return rate), maps required backend guarantees (inventory sync and payment tokenization), and specifies privacy/provenance controls tied to SynthID detection for generated audio. (gapinc.com, blog.google)
