Phonic launches speech-to-speech model Phonic-1 with $27.5M Series A

On October 29, voice AI startup Phonic released Phonic-1, a speech-to-speech model, and announced a $27.5 million Series A. Phonic-1 is positioned as the most advanced speech-to-speech model for agentic tasks; the company says traditional models in this class sacrifice intelligence for low latency, whereas Phonic-1 runs in hard real-world phone calls across six continents. Training was supported by Together AI.

Confirmed

• $27.5M Series A led by First Round, with Lux Capital, Felicis, and Definition participating
• Angel investors include Modal CEO Erik Bernhardsson and the Hugging Face CEO (per a post relayed by @graceisford), among others
• The team of 10 comes from MIT, Stanford, and CMU, including international olympiad medalists and repeat founders
• Team philosophy: "building at the frontier requires going deep with customers" — they believe voice's hardest problems only surface in real calls, so they rebuilt every layer of the stack
• Per the company, monthly call volume has grown substantially over the past year

Why it matters

• Phonic is honing its model on real-world calls as its core path, diving straight into the fiercely competitive frontier of agentic voice tasks; the funding should accelerate competition in this space.

2026-10-29 ~ 2026-10-30 · 6 related posts

• [source] Phonic launches speech-to-speech model Phonic-1, raises $27.5M Series A — togethercompute · 2026-10-29
• Voice startup Phonic raises $27.5M Series A as monthly call minutes spike — graceisford · 2026-10-30
• Phonic-1 comes from a 10-person MIT/Stanford/CMU team rebuilding the voice stack — graceisford · 2026-10-30

3 near-duplicate retellings: charles_irl · graceisford · rohanpaul_ai