Shunya Labs logo

Shunya Labs models

Voice AI company that builds speech recognition, speech synthesis and translation models aimed at CPU and on-premises deployment.

Gurugram, Indiashunyalabs.ai

Releases
1
Open weights
1
Latest
Jul 24, 2025

July 2025

1 release
  1. SpeechOpen weights

    Pingala V1

    Pingala V1 is Shunya Labs' speech recognition model designed to run on ordinary CPUs, released as a multilingual universal checkpoint and an English verbatim checkpoint.

    Modalities
    audio → text
    Parameters
    0.8B
    License
    Shunya Labs RAIL-M
    At launch
    Open weights, API, Enterprise
    • Shunya Labs reported an average word error rate of 2.94% across eight benchmarks at launch and later listed 3.10% composite WER on the Open ASR leaderboard set
    • The company said it ran on CPUs without GPUs, with latency under 250 ms, and offered it through an API, Docker and edge packages
    • Weights were published on Hugging Face behind an access form under a use-restricted RAIL-M license that allowed free use up to 10,000 audio hours a month
    • The company said it first built the model to improve transcription for mental health care
    Source: Shunya Labs Launches Pingala V1: Top-Ranked CPU-Based ASR Model Setting Performance Records

Cite and reuse this data

These release records are free to reuse under CC BY 4.0, including commercially. Credit The AI Engineering and link to this page's URL.

Dates are the labs' official announcement dates, and every record links to its source. List fields in the CSV are separated by " | ".

Citation

The AI Engineering Editors. "Shunya Labs models: release history and timeline." The AI Engineering, updated Jul 24, 2025. https://theaiengineering.com/releases/shunya-labs

HTML link

<a href="https://theaiengineering.com/releases/shunya-labs">Shunya Labs models: release history and timeline</a> (The AI Engineering)
Full dataset
1,250 releases from 73 labs
Coverage
to
Updated
License
CC BY 4.0
Publisher
The AI Engineering

Other labs