Gemini adoption surges in Southeast Asia with local-language, mobile-first use and agent rollout
Key Points
- Active users doubled year-over-year
- ~70% of prompts in native languages
- Gemini Spark launching in local languages
Summary
Gemini adoption in Southeast Asia has accelerated: active users in the region more than doubled year-over-year. Growth is driven by young, mobile-first users who prefer native languages and multimodal inputs. Key creative and productivity metrics include ~70% of prompts in local languages, ≈75% of requests from mobile, 5 billion images generated with Nano Banana, and ~1 million songs with Lyria 3. Google is rolling out Gemini Spark (proactive AI agents) in local languages to Gemini Advanced subscribers; Spark integrates with Workspace (Gmail, Docs, Slides) and can run while devices are locked.
Key Points
- Active users doubled year-over-year in Southeast Asia.
- ~70% of prompts use native languages (Vietnam 89%, Thailand 87%, Indonesia 84%); SEA-HELM ranks Gemini best for regional languages.
- ≈75% of requests originate on mobile; >40% of prompts use photos/video/voice; voice-only ≈10%.
- Heavy creative usage: 5B images (Nano Banana) and ~1M songs (Lyria 3) generated in the region.
- Gemini Spark (background agents) launching in local languages for Advanced subscribers; deep Workspace integration and background task execution.
Engineering implications
- Prioritize multilingual datasets and regional benchmarks (SEA-HELM style) to maintain local fluency.
- Optimize mobile latency and support robust multimodal pipelines (speech, camera, images, video).
- Build secure background/agent execution models with strong auth, privacy controls, and locked-device constraints.
- Instrument per-country usage and prompt patterns (slang, code-switching) to fine-tune models and UX.