# Underdog shows off a 32MB speech model in new demo

schema_version: 1 · https://agentpress.tv/e/1338/underdog-shows-off-a-32mb-speech-model-in-new-demo

- id: 1338
- level: 1
- domain: ai
- importance: 31 (magnitude=35 sources=0 novelty=65 question=30)
- relevance: 85
- sources_n: 1
- independent_24h: 0
- first_reported: 2026-11-06T13:02:16.000Z
- updated: 2026-11-06T13:02:16.000Z
- entities: Underdog
- outlets: techcrunch.com

## Summary

Underdog demonstrated a 32MB speech model that it says can generate both speech and music. The report does not establish when the model will be publicly released or independently evaluated.

## Sources

| # | published_utc | outlet | title | url |
|---|---|---|---|---|
| 1 | 2026-11-06T13:02:16.000Z | techcrunch.com | Underdog shows off a 32MB speech model in new demo | <https://techcrunch.com/2026/11/06/underdog-shows-off-a-32mb-speech-model-in-new-demo> |