Neywa Labs
Neywa Labs builds infrastructure that makes frontier open AI models run natively on personal devices, especially Apple Silicon. It develops open-source engines and native applications so people can use capable AI privately, offline, and without depending on the cloud for every interaction.
At a Glance
- Developers and AI builders
- Model labs and open-model teams
- Organizations needing private or offline AI
- Users and teams running Apple Silicon Macs
- +1 more
AI Tools by Neywa Labs
(1)MLX-Audio
Audio Models for Apple Silicon
Discussions
No discussions yet
Be the first to start a discussion about Neywa Labs
Latest News
Nemotron Transcribe on Nativ: speech understanding on Apple Silicon, including transcription, speakers, translation, and spoken questions; Neywa reported 33.8× real-time processing on an M3 Ultra.
MiniCPM-OCR on MLX-VLM: OpenBMB's MiniCPM-OCR 1.0 and Jev became available on MLX-VLM with 16 public checkpoints for local document parsing and evaluation.
Neywa Labs came out of stealth and introduced its local-AI infrastructure for building and deploying AI on devices people already own.
Docling × MLX-VLM case study: IBM Research's document AI ran on Apple Silicon through MLX-VLM, with a reported 9.2× improvement for a four-page batch on an M3 Max.
Products & Services
An open-source macOS application for running open AI models locally on Apple Silicon. It provides a model library, local chat and multimodal workflows, agent and developer-tool connections through standard APIs, and offline processing without per-token cloud bills or sending content to a cloud service.
Neywa's local-AI engine and infrastructure layer for building AI products that run on personal hardware. It combines model support, inference, quantization, memory optimization, and device-aware performance, with MLX-VLM and MLX-Audio as highlighted libraries.
An open-source package for inference and fine-tuning of vision-language models on Mac using Apple's MLX framework. Neywa's company page describes it as supporting 5,000+ models and reports more than 5.1 million downloads at the time of research.
An open-source Apple-Silicon library for text-to-speech, speech-to-text, speech-to-speech, and music generation. It supports multiple model architectures, multilingual audio, voice customization and cloning, streaming, quantization, an interactive web interface, an OpenAI-compatible REST API, and Swift integration.
Market Position
Neywa positions itself as an end-to-end local-AI stack rather than only a model wrapper: its Engine provides open-source infrastructure for model support, inference, quantization, and device optimization, while Nativ turns those capabilities into a Mac application. Its differentiation is native Apple-Silicon execution, open development, offline/privacy-preserving operation, and avoiding per-token cloud costs.
Leadership
Founders
Prince Canuma
Co-founder and CEO. He spent years building open-source infrastructure that enables developers worldwide to run state-of-the-art AI models on hardware they already own; he is also associated with the Blaizzy GitHub account and MLX projects.
Sebastien Michaud
Co-founder and COO. His career is described by Neywa as focused on commercializing unreleased solutions to critical problems; at Neywa he focuses on removing the tradeoff between access and user agency in AI.
Lucas Newman
Co-founder and CTO / Head of Research. Neywa describes his career as solving difficult engineering problems that bring zero-to-one innovations to market; he leads research into what is possible in local AI.
Executive Team
Prince Canuma
Co-founder, CEO
Open-source AI infrastructure builder associated with the Blaizzy GitHub account and projects for running advanced models on owned hardware.
Sebastien Michaud
Co-founder, COO
Commercialization and operations leader focused on turning unreleased technical solutions into products and improving user agency over AI.
Founding Story
Neywa was created around the view that intelligence should be closer to people. Its founders argue that modern AI is usually located in data centers, forcing users to send private context to external providers and incur connectivity, latency, usage, and recurring inference costs. The company was publicly introduced from stealth in October 2026 to build infrastructure and products that let people develop and deploy AI on devices they already own.
Business Model
Revenue Model
The currently listed Services are free. Neywa distributes open-source software and local applications, while its site presents Nativ Pro as a future product and says paid-plan terms, pricing, billing, and cancellation details will be shown before purchase if paid plans are offered.
Pricing Tiers
Neywa's Terms of Use state that its Services are currently free; Nativ and Engine are open-source software released under the MIT License.
A planned product for scalable AI products and agents, with a workspace dashboard for token usage and model-performance tracking; no price is listed.
Target Markets
- Developers and AI builders
- Model labs and open-model teams
- Organizations needing private or offline AI
- Users and teams running Apple Silicon Macs
- Companies building document, vision, speech, video, and multimodal AI products
- Private local chat and assistant workflows
- Offline AI for users with sensitive or confidential information
- Building and shipping local AI features and applications
- Document understanding and OCR
- Local speech transcription, translation, speaker attribution, and spoken-question answering
- Text-to-speech, voice cloning, and conversational audio generation
- IBM Research / Docling
- OpenBMB
- Sinners
- En Busca del Fuego