Forward Future Tools Library

Sesame CSM 1B
Sesame CSM 1B generates English conversational speech from text and audio inputs, making it useful for developers building voice assistants and interactive agents.
Try Sesame CSM 1B →
huggingface.co·From $7/M characters·Checked 2026-08-28


![huggingchat/chat-ui · [NEW] Assistants](/_next/image?url=https%3A%2F%2Fmch9boduqwxvwzvs.public.blob.vercel-storage.com%2Ftools%2Fenrichment-images%2Fsesame-csm-1b-300103debb8a.png&w=3840&q=75)


›What is Sesame CSM 1B?
CSM 1B is Sesame's conversational speech generation model. It generates RVQ audio codes from text and audio inputs, using a Llama backbone and Mimi audio decoder. The model is available through Hugging Face and has an interactive demo.
›What are the pros and cons of Sesame CSM 1B?
Strengths
Apache License 2.0 supports open-source use and deployment
Native Hugging Face Transformers integration removes the need for custom inference code
Handles text and audio inputs for dialogue-oriented speech generation
Offers both conversational and read-speech voice styles
Trade-offs
The hosted API output is English speech, limiting multilingual use
The model has a 4K context window
The listed API route uses a single provider and has no model-level failover target
›What are Sesame CSM 1B’s key features?
Generates conversational English speech from text input
Accepts audio inputs and prior turns as conversational context
Supports voice options for conversational and read-speech styles
Uses a Llama backbone with a Mimi audio decoder
Loads natively through Hugging Face Transformers 4.52.1 and later
Available through an interactive Hugging Face demo
›What are the best use cases for Sesame CSM 1B?
Building voice assistants
Creating interactive voice agents
Generating multi-turn conversational audio
Testing speech generation through the Hugging Face demo
›What is the pricing for Sesame CSM 1B?
| Plan | Price | Details |
|---|---|---|
| Standard | $7/M characters | Usage-based API access for CSM 1B speech generation, billed by characters. |
The listed provider is DeepInfra through OpenRouter or Opper. Opper lists the equivalent rate as $7.00 / 1M chars.
Checked 2026-08-28 · source
›Who is Sesame CSM 1B best for?
developersA strong fit for developers building dialogue-oriented voice assistants or interactive agents.
AI researchersNative Transformers support and an Apache 2.0 license make it practical for experimentation and local deployment.
Not for
- Teams that need speech output in languages other than English should look for a multilingual model.
- Projects requiring a long conversational context may outgrow the 4K context window.
- Production systems that require provider redundancy should avoid relying on the single listed route.