Nvidia PersonaPlex 7B
Ranked 3 of 6 voice AI models by recommenders
Recommended for natural full-duplex speech conversation and a permissive license, though critics find it unintelligent and hard to keep stable without tool calling.
Recommended for natural full-duplex speech conversation and a permissive license, though critics find it unintelligent and hard to keep stable without tool calling.
22 people gave an opinion
- recommend it
- 14
- criticize it
- 8
- approval share
- 64%
Recommended in 4 of 13 months
- November 2025: not recommended
- December 2025: not recommended
- January 2026: recommended
- February 2026: not recommended
- March 2026: recommended
- April 2026: not recommended
- May 2026: recommended
- June 2026: not recommended
- July 2026: recommended
- August 2026: not recommended
- September 2026: not recommended
- October 2026: not recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 14 people recommend itNeeds 10 or more, met.
- 1.7 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 4 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Natural full-duplex conversation flow
“PersonaPlex stands out by tightly integrating voice, prosody, and real-time interaction”
Comment from 22 Jan 2026 in “Nvidia: Natural Conversational AI with Any Role and Voice” “Seems like only Sesame and NVDIA so far have nailed the natural conversation flow”
Comment from 27 Jan 2026 in “Show HN: LemonSlice – Give your voice agents a face” “the NVIDIA Personaplex demos are very impressive in this regard”
Comment from 28 Jan 2026 in “Show HN: LemonSlice – Give your voice agents a face” - Comment from 5 Mar 2026 in “Nvidia PersonaPlex 7B on Apple Silicon: Full-Duplex Speech-to-Speech in Swift”
Single model handles voice and prosody
“Another good follow-up would be to try PersonaPlex, Nvidia's new model that would completely replace this architecture with a single model that does everything”
Comment from 2 Mar 2026 in “Show HN: I built a sub-500ms latency voice agent from scratch” “PersonaPlex stands out by tightly integrating voice, prosody, and real-time interaction”
Comment from 22 Jan 2026 in “Nvidia: Natural Conversational AI with Any Role and Voice” “the go to solution is PersonaPlex from Nvidia based upon Moshi”
Comment from 8 Jul 2026 in “GPT‑Live” Promising and worth trying
“It doesn't say useful stuff and there's no clear path to plugging it into smarter models, but it's worth experiencing.”
Comment from 5 Mar 2026 in “Nvidia PersonaPlex 7B on Apple Silicon: Full-Duplex Speech-to-Speech in Swift” “PersonaPlex is still interesting although it was looking like we would need to fine tune it”
Comment from 8 Jul 2026 in “GPT‑Live” “PersonaPlex totally seems like the future.”
Comment from 4 May 2026 in “How OpenAI delivers low-latency voice AI at scale”
What people warn about
Each mark is one comment
Limited intelligence without tool calls
“without tool calling this is basically a fast parrot”
Comment from 5 Mar 2026 in “Nvidia PersonaPlex 7B on Apple Silicon: Full-Duplex Speech-to-Speech in Swift” “7b can't be very intelligent on its own, and for anything useful, you'd need tool-calls”
Comment from 5 Mar 2026 in “Nvidia PersonaPlex 7B on Apple Silicon: Full-Duplex Speech-to-Speech in Swift” “its quality is quite far from GPT-Live-1, esp in terms of intelligence”
Comment from 15 Sep 2026 in “Show HN: Nari Qwen3-TTS and Qwen3-ASR – High accuracy, low latency and cost” - Comment from 16 Feb 2026 in “Why Audio Is the One Area Small Labs Are Winning”
Choppy, slow, or inaccurate performance
“this full duplex architecture is still a bit off in terms of giving you good accuracy/performance”
Comment from 5 Mar 2026 in “Nvidia PersonaPlex 7B on Apple Silicon: Full-Duplex Speech-to-Speech in Swift” “it was absolutely terrible. Around 10 seconds for each reply, and even then it was saying something totally unrelated.”
Comment from 5 Mar 2026 in “Nvidia PersonaPlex 7B on Apple Silicon: Full-Duplex Speech-to-Speech in Swift” “couldn't get even PersonaPlex to run properly (without choppiness/latency)”
Comment from 28 Jan 2026 in “Show HN: LemonSlice – Give your voice agents a face” No text output alongside audio
“it lacks the ability to have a silent text stream output”
Comment from 5 Mar 2026 in “Nvidia PersonaPlex 7B on Apple Silicon: Full-Duplex Speech-to-Speech in Swift” “If the model cannot emit both text and audio at the same time, to enable tools, it’s not really useful at all for voice agents.”
Comment from 5 Mar 2026 in “Nvidia PersonaPlex 7B on Apple Silicon: Full-Duplex Speech-to-Speech in Swift” “we have to also separately transcribe because it does not also provide what the person said”
Comment from 5 Mar 2026 in “Nvidia PersonaPlex 7B on Apple Silicon: Full-Duplex Speech-to-Speech in Swift” Gets stuck, stutters, or rambles
“it seems to work in a rough way but only if I turn the lora scaling factor up to 5 and that generally screws it up in other ways”
Comment from 5 Mar 2026 in “Nvidia PersonaPlex 7B on Apple Silicon: Full-Duplex Speech-to-Speech in Swift” “it's really easy to get PersonaPlex stuck in a death spiral - talking to itself, stuttering and descending deeper and deeper into total nonsense”
Comment from 5 Mar 2026 in “Nvidia PersonaPlex 7B on Apple Silicon: Full-Duplex Speech-to-Speech in Swift”
Other voice AI models people recommend
The four most recommended other voice AI models, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| GPT-Live | 72 | 48 | 60% | 4 | — | — |
| Gemini Live | 50 | 23 | 68% | 9 | — | Approved |
| Nvidia PersonaPlex 7BThis page | 14 | 8 | 64% | 4 | — | — |
| gpt-realtime | 6 | 6 | 50% | 4 | — | — |
| Sesame voice demo | 5 | 1 | 83% | 3 | — | — |
What people said
All 27 comments that mention it, newest first. Each links to the original comment.
“its quality is quite far from GPT-Live-1, esp in terms of intelligence”
“the go to solution is PersonaPlex from Nvidia based upon Moshi”
“PersonaPlex is still interesting although it was looking like we would need to fine tune it”
“PersonaPlex totally seems like the future.”
“we have to also separately transcribe because it does not also provide what the person said”
“It doesn't say useful stuff and there's no clear path to plugging it into smarter models, but it's worth experiencing.”
“without tool calling this is basically a fast parrot”
“7b can't be very intelligent on its own, and for anything useful, you'd need tool-calls”
“it was absolutely terrible. Around 10 seconds for each reply, and even then it was saying something totally unrelated.”
“If the model cannot emit both text and audio at the same time, to enable tools, it’s not really useful at all for voice agents.”
“it works well for me to toggle lights on and off”
“it seems to work in a rough way but only if I turn the lora scaling factor up to 5 and that generally screws it up in other ways”
“overlapping realistic full duplex as moshi/personaplex”
“it's really easy to get PersonaPlex stuck in a death spiral - talking to itself, stuttering and descending deeper and deeper into total nonsense”
“this full duplex architecture is still a bit off in terms of giving you good accuracy/performance”
“it lacks the ability to have a silent text stream output”
“it does not allow any kind of interactive conversation”
“It's not surprising that people have little interest in talking to AI if they're being lied to.”
“It just about works for our current use case but can't comprehend the concept of an outgoing call”
“If you're of that opinion, you'll enjoy the new stuff coming out from nvidia”
“Another good follow-up would be to try PersonaPlex, Nvidia's new model that would completely replace this architecture with a single model that does everything”
“in my test it was oddly dumb and unable to follow instructions”
“couldn't get even PersonaPlex to run properly (without choppiness/latency)”
“the NVIDIA Personaplex demos are very impressive in this regard”
“Seems like only Sesame and NVDIA so far have nailed the natural conversation flow”
“It has dual channel input / output and a very permissible license”
“PersonaPlex stands out by tightly integrating voice, prosody, and real-time interaction”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.