🇺🇸 USA · Google Project Astra
Status: 🟩 COMPLETE 🟦 LIVING Last updated: 2026-06-26 Plain-English tagline: Google’s universal AI assistant — point your phone camera at something, talk to Astra about what it sees. The visual / conversational future of Gemini, rolling out gradually.
Front-matter facts
| Field | Value |
|---|---|
| Vendor | Google DeepMind (Mountain View, USA + London, UK) |
| Country / origin | 🇺🇸 USA + 🇬🇧 UK (DeepMind) |
| Recommended for Australian users? | ✅ Yes — rolling out in AUS via Google AI Pro / Ultra subscriptions and Pixel devices |
| Privacy summary | Camera + voice data processed via Gemini infrastructure; Gemini Apps Activity opt-out applies; sensitive imagery processed with usual AI safety considerations |
| Free tier | Limited preview access |
| Paid tiers | Bundled with Google AI Pro / Ultra; expanding to broader Pixel + Workspace users |
| First released | Announced May 2024 at Google I/O; preview rollout 2024-25; broader availability 2025-26 |
| Last reviewed | 2026-06-26 |
| Official site | https://deepmind.google/technologies/project-astra/ |
What it is
Project Astra is Google’s universal AI assistant — built around real-time vision + conversation. The pitch: point your phone camera at anything (a plant, a book, a problem on a whiteboard, a building, your fridge contents), and have a fluid spoken conversation with Astra about what it sees.
Key capabilities:
- Real-time camera vision — Astra sees through your phone / glasses camera live
- Continuous conversation — multi-turn dialogue, remembering context
- Memory across sessions — “remember where I left my keys” (then show it via camera)
- Multimodal reasoning — text + image + audio + video simultaneously
- Tool use — can interact with apps / Google Maps / Calendar
- Glasses-form-factor potential — Astra is the AI capability behind Google’s smart-glasses experiments
- Low latency — responses feel conversational, not chatbot-paced
Project Astra is the spiritual successor to Google Lens + Google Assistant + Gemini Live combined — a unified always-available visual / conversational AI.
Initial rollouts:
- Pixel devices first
- Google AI Ultra subscribers (preview access)
- Workspace Enterprise customers experimenting
- Smart-glasses hardware partners (Samsung, others)
What you’d use it for
- “What is this thing?” point-camera-and-ask scenarios
- Translation — point at a foreign menu, hear translation
- Cooking — show fridge contents, get recipe suggestions
- Repair / DIY — show a broken appliance, get diagnosis
- Plants / animals / objects — identification with conversation
- Whiteboard / problem-solving — show maths or diagram, work through it
- Travel — point at landmarks, hear history
- Memory aid — Astra remembers things across sessions (“where did I put X?”)
- Accessibility — verbal descriptions for vision-impaired users
How to get from Australia
- Project Astra access varies — currently most-available via:
- Google AI Ultra subscription (US$249.99/mo) — preview access
- Pixel 8+ / 9+ / 10+ devices (rolling out)
- Workspace Enterprise preview programs
- Gemini Live (the broader voice mode) is available on Google AI Pro and free tier; some Astra capabilities flow through Gemini Live
- Full Project Astra (especially camera-based always-on) requires compatible hardware + subscription
- Smart-glasses-form-factor Astra requires specific hardware (rolling out via partners)
What it costs
Currently via subscription / hardware
- Google AI Ultra: US390 AUD) — best Astra access
- Pixel devices: hardware cost
- Smart glasses partners: hardware cost (varies)
Future (likely)
- Broader Pixel inclusion
- Google AI Pro tier expansion
- Workspace Enterprise inclusion
How it compares to alternatives
| Capability | Project Astra | OpenAI Advanced Voice + Vision | Anthropic Claude voice + vision | Apple Intelligence Visual |
|---|---|---|---|---|
| Camera-live conversation | Best (purpose-built) | Yes (Advanced Voice with video) | Rolling out | Visual Intelligence (iPhone) |
| Smart-glasses readiness | Strongest | Limited | Limited | Apple Vision Pro |
| Cross-app actions | Yes | Yes (Operator) | Computer Use | Yes (Apple ecosystem) |
| Memory across sessions | Yes | Yes (ChatGPT Memory) | Yes (Claude Memory) | Yes (Apple Intelligence) |
| AUS availability | Rolling out | Available now | Available now | Available now |
| Best for | Pixel users + smart-glasses future | iPhone + ChatGPT users | Anthropic ecosystem | Apple users |
Project Astra is the most-ambitious visual-AI vision of the major Western labs, but it’s still rolling out — competitors (OpenAI Advanced Voice with vision, Apple Visual Intelligence) ship more broadly today.
Privacy / data handling
- Camera + voice data processed via Gemini infrastructure
- Gemini Apps Activity opt-out applies (myactivity.google.com/product/gemini)
- For Workspace tier: no training, tenant-isolated
- Camera-always-on raises real privacy questions — Google has been cautious with rollout because of this
- For smart-glasses-form-factor, additional considerations about consent of people captured
Recent changes
- 2026: Broader rollout including AUS preview
- 2025: Pixel 9 / 10 integration
- May 2024: Project Astra announced at Google I/O
Gotchas
- Currently preview / limited access — full Astra not yet broadly available
- Hardware-dependent — best on Pixel / smart glasses
- Privacy considerations with camera-always-on
- For “I want to use it today” in AUS, Gemini Live (subset of Astra capabilities) is more accessible than full Astra
- Apple Visual Intelligence + Advanced Voice with Vision are competing today; Astra leads on glasses-form-factor vision
See also
- Gemini 🟩 🟦
- Vertex AI 🟩 🟦
- Apple Intelligence (Visual Intelligence) 🟩 🟦
- ChatGPT (Advanced Voice) 🟩 🟦
- Claude in Chrome 🟩 🟦
- Galaxy AI 🟩 🟦
- Multimodal (vision, audio) 🟩 🟦
- Project Mariner 🟥 — Google’s browser-agent sister