VQV Signal

OPEN SOURCE WATCH TECHNICAL

Haru, an open 4B full-duplex voice model, gets a one-click local launcher

According to a post on X by @cocktailpeanut, a one-click launcher now lets people run Haru on their own computer, and the author pitched it as an AI friend like the one in the movie Her. The post describes Haru as a 4B full-duplex speech model, open-sourced under Apache 2.0, and says its makers trained it on 40 million hours of audio. The launcher thread says it was tested on a Mac, but the snippet cuts off before the hardware details. The post had about 770 likes and 800 bookmarks at the time of the snippet.

Source: Scobleizer News Pulse (X) · x.com Found via Robert Scoble’s Scobleizer News Pulse · a post on X, not independently verified Published 2026-11-13T17:45:00+00:00 Detected 2026-11-13T22:26:49+00:00
View original source

According to a post on X by @cocktailpeanut, a one-click launcher now lets people run Haru on their own computer, and the author pitched it as an AI friend like the one in the movie Her. The post describes Haru as a 4B full-duplex speech model, open-sourced under Apache 2.0, and says its makers trained it on 40 million hours of audio. The launcher thread says it was tested on a Mac, but the snippet cuts off before the hardware details. The post had about 770 likes and 800 bookmarks at the time of the snippet.

AI-assisted summary based on the listed source.

cocktail peanut released a one-click launcher to run Haru on your own computer and pitched it as an AI friend like the one in the movie Her. Haru is a 4B full-duplex speech model, open-sourced under Apache 2.0 and trained on 40 million hours of audio according to its makers, whose announcement on Thursday evening...

Developers and users who want to experiment with full-duplex voice AI locally may find it easier to start, since the launcher is aimed at running the model on a personal computer. The Apache 2.0 license is permissive, which matters for anyone considering building on the model. Watch for independent testing of the hardware requirements and the training claims, which this source does not confirm.

Signal Strength 91% Technical label WATCH Public Interest 47 Category OPEN SOURCE Reader Depth TECHNICAL

Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.

Public Interest components
Recognizable Entity Score 0 Practical Impact Score 82 Novelty Interest Score 94 Consequence Score 18 Curiosity Score 32 Shareability Score 60

VQV surfaced this signal because it is recent, relevant to AI Voice, connected to Scobleizer News Pulse (X).