10-person startup Phonic launches Phonic-1 speech model, raises $27.5M Series A
rohanpaul_ai · x · 2026-10-30
Phonic, a 10-person startup, released Phonic-1, a speech-to-speech model that can run tool lookups and bookings while the caller is still mid-sentence, alongside a $27.5M Series A led by First Round ($31.5M total).
Key numbers:
- Tops Artificial Analysis's speech-to-speech index at 87.1, edging xAI's Grok Voice 3.0 (86.9); completes 79.9% of customer-service tasks with 0.61s time-to-first-audio.
- In demos it finished 5 lookups mid-speech and replied 1.1s after the caller stopped, vs 15.7s for OpenAI's GPT-Live.
- Scores 90.2 on Cekura's call suite, beating an 82.9% transcribe-think-speak cascade — though Cekura's public board lists Phonic v1 at just 68.3%.
- Podonos blind tests: listeners prefer Phonic-1 over GPT-Live-1 (61.5%), Grok Voice 3 (66.5%) and Eleven v4 Turbo (78%), but still prefer human voice actors (57%).
- Already powers Ashby's AI job interviews across 25 countries.
More from Venture
- Twilio stock plummets; CEO pitches it as key AI voice and shopping-bot infrastructure — TiernanRayTech · 2026-10-30
- North Korea-linked hackers drain $6.36B from Binance in 20 minutes, largest crypto theft ever — kimmonismus · 2026-10-30
- AI Stocks Wobble but Broader Market Holds On, NYT Reports — nordicinst · 2026-10-30
- He runs his startup with 11 Claude agents — 31.4B tokens and ~$16.8k of API cost, prompt open sourced — Mental_Order5696 · 2026-10-30
- Outrank Marketplace pays your idle domains by DR: up to $1,200/month — tibo_maker · 2026-10-30
- SanDisk: ~50% of FY27 NAND supply locked in multi-year deals, datacenter bit demand +40% in 2027 — tengyanAI · 2026-10-30