Sesame
Sesame is a voice-first AI assistant app for iOS and Android with lifelike agents like Maya and Miles, built on its own conversational speech model
Overview
Sesame is a voice-first AI assistant from the company led by Oculus co-founder Brendan Iribe. Its agents are "everyday collaborators who help you do more, talk things through, and follow your curiosity". You talk to them out loud in the Sesame app on iOS and Android, or in the browser research preview, and they remember earlier conversations. Sesame became known in February 2025 for its Maya and Miles voice demo, which many testers found the most human-sounding voice AI yet, and it released its Conversational Speech Model (CSM-1B) as open weights under Apache 2.0. Smart glasses with the assistant built in are announced but not shipped.
The Verdict
Who Should Use Sesame?
Best For
- People who prefer talking to an assistant rather than typing
- Anyone curious how natural voice AI can sound today
- Developers who want an open speech-generation model to experiment with (CSM-1B)
Not Ideal For
- Businesses that need phone, CRM or call-analytics integrations
- Non-English speakers: the app is optimized for English
- Users who want deep research, files or coding help in the same app
What's Great
- Very natural, expressive voices; testers compare it favorably with ChatGPT voice mode
- Persistent memory across conversations in the app
- Free to use today
- Open CSM-1B model on GitHub and Hugging Face under Apache 2.0
Watch Out For
- Conversations can feel shallow and repeat sentences, per user reviews
- English-first; multilingual support is still in progress
- No business features such as telephony or CRM hooks
- No public pricing, so the free period may change
- The open CSM repository has not been updated since May 2025, and the weights are gated
Pricing
View all features & details
Key Features
- Real-time voice conversation with several agent personas, including Maya and Miles
- Memory of past conversations
- Mobile apps plus a browser research preview
- CSM-1B speech model with a Llama-3.2-1B backbone
- Glasses with built-in assistant announced as coming soon
Use Cases
- Talking through ideas or plans hands-free
- Casual questions and curiosity-driven chats
- Research and prototyping with an open speech model
The iOS app launched in May 2026. Sources: Enterprise DNA · Crossing the uncanny valley of voice (Feb 2025) · Research preview
How It Compares
| Feature | Sesame | ChatGPT |
|---|---|---|
| Main mode | Voice-first | Text first, with voice mode |
| Price | Free (no paid plans yet) | Free, Go $8/mo, Plus $20/mo |
| Open model | CSM-1B (Apache 2.0) | No |
| Business integrations | None | Yes |
Sources: Sesame · ChatGPT page
Is this your tool? Add a badge