Sesame CSM
Sesame class A research
Sesame's open-weights (Apache-2.0) 1B conversational speech base model; English-focused, produces varied voices from context audio, no packaged clone.
Layer 1 — factual axes
| API & price | self-host only (open weights) | official | source ↗ 2026-07-04 |
|---|---|---|---|
| Voices | any (design) | official | source ↗ 2026-07-04 |
| Voice cloning | no (contextual matching, not packaged clone) | official | source ↗ 2026-07-04 |
| Languages | 1 (English) | official | source ↗ 2026-07-04 |
| Weights | open · Apache-2.0 | official | source ↗ 2026-07-04 |
| Free tier | open self-host | official | source ↗ 2026-07-04 |
Layer 2 — output quality
No independent measure yet. We publish a quality figure only when a named external benchmark ranks this model — we don’t invent one.
Sesame’s open-weights (Apache-2.0) 1B conversational speech base model; English-focused, produces varied voices from context audio, no packaged clone.