Skip to content

Search stories, models, companies…Subscribe

Topic · Concept

Voice AI

Voice AI covers models that turn speech into text, text into speech or speech into speech, and the voice assistants and agents built on them. This page tracks the speech models labs release, what they cost per minute, which ones have open weights, and the rules on AI-generated and cloned voices.

7 stories · updated Oct 29

Follow via RSS Get Voice AI alerts

Key facts

Checked Oct 25

Whisper speech recognition paper: "Robust Speech Recognition via Large-Scale Weak Supervision," Alec Radford et al. (OpenAI); arXiv 2212.04356, December 6, 2022; scaled to 680,000 hours of multilingual and multitask supervision
AI voices in US robocalls: FCC Declaratory Ruling announced February 8, 2024, adopted unanimously: calls made with AI-generated voices are "artificial" under the TCPA; effective immediately
GPT-Live-1 in the API: September 10, 2026: OpenAI's full-duplex voice model, $0.05 per minute for the front-end voice layer
GPT-Live-1 billing: $0.05 per minute, billed per second; backend model and tool usage billed separately
GPT-Live-1.1 in ChatGPT: October 15, 2026: rolling out to all users in ChatGPT, Codex and dots, with 13 new voices; "Coming soon to the API" (not in OpenAI's API model list on October 25, 2026)
ChatGPT Voice limits: Live runs GPT-Live-1.1 or GPT-Live-1.1 Quip by plan, per rolling 24 hours: Free, limited Quip; Go, 3 hours of Quip; Plus, 3 hours of GPT-Live-1.1; Pro ($200/month), unlimited GPT-Live-1.1 (with a footnote)
Grok Voice 3.0: Launched October 14, 2026, at $0.08 per minute ($4.80 per hour); xAI calls it "the world's most intelligent speech-to-speech model"
Grok voice API pricing: grok-voice-3.0: $0.08 per minute, $4.80 per hour, $0.004 per text input; grok-voice-2.0 listed as deprecated at the same price
Voxtral 3 models and prices: Voxtral Small Transcribe 3 $0.004/min; Mini Transcribe 3 (4B parameters) $0.0035/min; Mini Transcribe Realtime 3 $0.008/min; Mini TTS 3 $13 per million characters; voice agent API $0.04/min with Mistral Small 4, $0.08/min with a larger model
Voxtral 3 release and status: Released October 20, 2026; labeled "Public Preview" and "Premier" in Mistral's docs
Voxtral open weights: No Voxtral 3 repository on Mistral's Hugging Face account on October 25, 2026; it lists Voxtral Mini 3B and Small 24B (2507), Mini 4B Realtime (2602, plus an Arabic variant) and 4B TTS (2603)
Voxtral Mini 3B (2507): Apache 2.0; built on Ministral 3B; 8 languages; 32k-token context
Block Dessa: Research preview of a full-duplex speech-to-speech model that Block says it trained from scratch on human conversation, with "no text, no pretrained models," on one Nvidia GB200 NVL72; runs locally on a MacBook Pro M4 Max or better; early access for a small group of AI safety researchers; "Launch links are coming soon"

Start here

Models · NewsBlock's Dessa: a full-duplex conversation model trained only on speech, with no text, released as a closed research previewBlock's Dessa is a speech-to-speech conversation model that Block says was trained from scratch on human conversation audio, with no text and no pretrained models. What that claim means, what Block has disclosed, and what it hasn't: size, license, benchmarks or a release date.Oct 26·4 min read

Models · NewsVoxtral 3: Mistral's new speech-to-text, text-to-speech and voice-agent SDK are API-only, unlike earlier Voxtral modelsVoxtral 3, released October 20, 2026, adds three Mistral transcription models, a text-to-speech model and a voice-agent SDK priced from $0.0035 a minute. Unlike earlier Voxtral releases, none of the Voxtral 3 models has open weights.Oct 24·5 min read

Products · NewsGPT-Live-1.1 and Grok Voice 3.0: what's new in ChatGPT Voice and xAI's voice APIOpenAI began rolling out GPT-Live-1.1 and a smaller GPT-Live-1.1 Quip in ChatGPT Voice on October 15, 2026, with 13 new voices and better accents. A day earlier, xAI released Grok Voice 3.0 in its Voice Agent API at an unchanged $0.08 a minute.Oct 16·5 min read

Latest on Voice AI

Open source · NewsNemotron 3.5 Transcribe: Nvidia's open speech recognition model covers 40 languages under OpenMDW, with about 4B active parametersNemotron 3.5 Transcribe, released by Nvidia on Hugging Face on October 28, 2026, is an open-weight speech recognition model for 40 languages under the OpenMDW-1.1 license. It activates about 4B parameters per token, runs on a single Nvidia GPU and handles recorded audio, not live streams. Nvidia's word error rates are vendor-reported.Oct 29·6 min read

Products · ExplainerAmazon Bee AI wearable: Bee Frontier costs $74.99, Bee Eclipse $99.99, plus what they record, the $11.99 Bee Pro plan and the consent questionAmazon's Bee AI wearable is back as the $74.99 Bee Frontier, with the $99.99 Bee Eclipse phone attachment coming soon and an $11.99-a-month Bee Pro plan. What they record, what Amazon says happens to the audio, where they're sold, and why recording-consent laws matter.Oct 27·8 min read

Open source · NewsHurston-1 TTS: an Apache 2.0 voice model for Southern American and African-accented English, built for phone callsHurston-1 is a 3-billion-parameter text-to-speech model from Atlanta's Destined AI, released on Hugging Face under Apache 2.0 and shared on Hacker News on October 25, 2026. It speaks Southern American and several African English accents, with an 8 kHz phone mode. What it was trained on, what its makers report, and its stated limits.Oct 27·5 min read

Models · NewsBlock's Dessa: a full-duplex conversation model trained only on speech, with no text, released as a closed research previewBlock's Dessa is a speech-to-speech conversation model that Block says was trained from scratch on human conversation audio, with no text and no pretrained models. What that claim means, what Block has disclosed, and what it hasn't: size, license, benchmarks or a release date.Oct 26·4 min read

Models · NewsVoxtral 3: Mistral's new speech-to-text, text-to-speech and voice-agent SDK are API-only, unlike earlier Voxtral modelsVoxtral 3, released October 20, 2026, adds three Mistral transcription models, a text-to-speech model and a voice-agent SDK priced from $0.0035 a minute. Unlike earlier Voxtral releases, none of the Voxtral 3 models has open weights.Oct 24·5 min read

Models · NewsKrea 3 and EmbeddingGemma 2 speech-to-text: two smaller AI releases worth knowingKrea 3, launched October 15, 2026, is an image model built for editing at scale at $0.03 to $0.06 an image. Separately, a developer turned Google's open EmbeddingGemma 2 into an on-device speech-to-text app. What each is, what's open, and what it costs.Oct 16·6 min read

Products · NewsGPT-Live-1.1 and Grok Voice 3.0: what's new in ChatGPT Voice and xAI's voice APIOpenAI began rolling out GPT-Live-1.1 and a smaller GPT-Live-1.1 Quip in ChatGPT Voice on October 15, 2026, with 13 new voices and better accents. A day earlier, xAI released Grok Voice 3.0 in its Voice Agent API at an unchanged $0.08 a minute.Oct 16·5 min read

Nothing in this format on this page — try another filter or the next page.