ZETIC
ZETIC builds Melange, an on-device AI deployment platform that automates model optimization, quantization, and NPU-aware compilation across Android and iOS, enabling mobile developers and enterprises to ship hardware-accelerated AI inference in hours rather than months without cloud dependency.
- Company typePrivate
- Founded2024
- HeadquartersSan Francisco, United States
- Headcount11–50
- GTM typeB2B
- OfferingSoftware
What ZETIC does
ZETIC is a developer infrastructure company building tools for on-device AI deployment on mobile hardware. Its core product, the Melange platform (commercially branded as MLange), automates model conversion, quantization, and NPU-aware compilation to enable AI models — primarily in ONNX and PyTorch Exported Program formats — to run directly on Android and iOS devices without cloud dependency. The platform abstracts hardware across Qualcomm HTP/DSP, MediaTek APU, Samsung Exynos DSP, and Apple Neural Engine (A11+) chipsets, with automated backend selection delivering up to 60x faster inference than CPU execution and 50% model size reduction through quantization. Deployment cycles that previously required months of manual NPU tuning are reduced to under 6 hours via a 3-step upload-benchmark-deploy workflow.
The product surface consists of the Melange web dashboard (mlange.zetic.ai) for model upload, benchmarking across 200+ real smartphone configurations, and credential provisioning; SDKs distributed through Maven Central (Android) and Swift Package Manager (iOS), including ZeticMLangeModel for general vision/audio inference, ZeticMLangeLLMModel for on-device LLM inference via a LLaMA.cpp backend (supporting GGUF quantization and multimodal embedding injection), and ZeticMLangeHFModel for direct Hugging Face integration; and proprietary runtimes shipped as commercial binaries. ZETIC has also extended into Embedded AI Solutions targeting MCUs through industrial computers and offering silicon benchmarking services for chipmakers.
The business operates on a tiered SaaS subscription model: a Free tier with limited hardware targets, Pro and Pro+ monthly tiers with expanded target limits (15 and 50 hardware targets) and bandwidth allocations (500GB and 2TB respectively) plus pay-as-you-go overages, and a quote-based Enterprise tier priced on annual download reach. Distribution is product-led through self-serve onboarding, Discord, GitHub, and Hugging Face community channels, supplemented by direct sales for enterprise contracts. The company was co-founded by Yeonseok Kim (former Qualcomm AI R&D Engineer) and Seongjun Kim, is incorporated in Delaware with a principal address in Dover, and received seed funding from Korea Investment Accelerator and TheVentures in May 2024.
ZETIC firmographics
Firmographics- Name
- ZETIC
- Legal name
- ZETIC.ai
- Website
- https://zetic.ai
- Company type
- Private
- Founded year
- 2024
- Operating status
- Operating
- Headcount range
- 11–50 employees
- Short description
- ZETIC builds Melange, an on-device AI deployment platform that automates model optimization, quantization, and NPU-aware compilation across Android and iOS, enabling mobile developers and enterprises to ship hardware-accelerated AI inference in hours rather than months without cloud dependency.
- Ownership category
- akta.pro rank
ZETIC industry classification
Industry- Product category
- On-Device AI Deployment Platform
- NAICS
- Software Publishers (513210)
- SIC
- Services-Prepackaged Software (7372)
- akta.pro primary industry
- Model Deployment, Serving & Inference Platforms (HDAAABAF)
- akta.pro secondary industries
- AI Compiler, Runtime & Kernel Optimization Software (CUDA/ROCm/XLA, graph compilers) (HDAAAAAI), Model Serving, Inference & Deployment Platforms (APIs, Edge/On-Prem) (HDAEANAC)
Keywords
Where ZETIC is headquartered
LocationHeadquarters
- HQ city
- San Francisco
- HQ country
- United States
- HQ region
- North America
Offices1 record
Markets served
ZETIC business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Personnel, Technology or R&D, Infrastructure, Marketing or Sales, Operations
Revenue model
- SaaS Subscription - Free Tier: Free tier providing access to the Melange Dashboard with limited features: supports Public Models only, limited to 4 flagship device optimization targets, with unlimited end user device installs.
- SaaS Subscription - Pro Tier: Monthly subscription including Public and Custom/User Models, up to 15 distinct hardware optimization targets, 500GB/month bandwidth allocation, with pay-as-you-go overages.
- SaaS Subscription - Pro+ Tier: Higher-tier monthly subscription with Extended Model Library, Custom/User Models, up to 50 hardware targets, 2TB/month bandwidth allocation, plus pay-as-you-go overages.
- SaaS Subscription - Enterprise Tier: Custom limits and features defined in a separate Enterprise Agreement. Usage monitored based on Estimated Reach (downloads) per annum. Includes advanced telemetry reports for Pro+ and Enterprise tiers.
- Overage Fees: Usage beyond tier allocations (bandwidth overages, additional benchmark requests) billed at per-GB or per-benchmark rates specified on the pricing page.
Pricing tiers
| Model | Billing | Price |
|---|---|---|
| Freemium | Monthly | Free tier: Public Models, 4 flagship device targets, unlimited installs |
| Subscription | Monthly | Pro tier: Public + Custom Models, 15 hardware targets, 500GB bandwidth |
| Subscription | Monthly | Pro+ tier: Extended Model Library, 50 hardware targets, 2TB bandwidth |
| Subscription | Annual | Enterprise tier: Custom limits, custom features, per-annum download-based scaling |
Go-to-market motion2 records
Distribution channels6 records
Marketing channels8 records
ZETIC product offering
Product offeringCore offering
ZETIC provides the Melange platform (also marketed as MLange), an on-device AI deployment service that automatically optimizes, converts, quantizes, and compiles AI models (ONNX and PyTorch Exported Program) for hardware-accelerated execution on mobile CPUs, GPUs, and NPUs. The platform benchmarks models across 200+ real smartphone devices and delivers production-ready Android and iOS SDKs in under six hours via a three-step workflow (upload, benchmark, deploy), replacing months of manual NPU tuning.
Product overview
ZETIC is building the infrastructure for mobile on-device AI. The product portfolio centers on the Melange platform (also marketed as MLange), which provides an end-to-end pipeline for optimizing and deploying AI models directly on mobile and embedded devices without cloud dependency. The core platform consists of the Melange web Dashboard for model upload, auto-compilation, and benchmark reporting; the Melange Runtime (including commercial Melange Pro Runtime and a Hugging Face Runtime Bundle) for on-device execution; and SDKs (ZeticMLangeModel for general vision/audio models, ZeticMLangeLLMModel for LLM inference, ZeticMLangeHFModel for direct Hugging Face integration) available for Android, iOS, and Flutter. A tiered subscription model (Free, Pro, Pro+, Enterprise) gates model support, hardware target limits, and bandwidth allocations. The platform also extends into Embedded AI Solutions for silicon benchmarking and industrial deployments. The company was founded by Yeonseok Kim (CEO, former Qualcomm AI R&D Engineer) and Seongjun Kim (Co-Founder & Business Lead).
Differentiator
Problem solved
Functional benefit
Brands
- Melange (also branded as MLange): On-device AI platform for deploying AI models directly on mobile devices, supporting CPU, GPU, and NPU acceleration across Android and iOS.
Products and services
- Melange On-device AI deployment platform that automatically optimizes, converts, and compiles AI models (ONNX and PyTorch Exported Program) for hardware-accelerated execution on mobile NPUs, GPUs, and CPUs. Provides a three-step workflow (upload, benchmark, deploy) with a 3-line SDK integration for Android (Kotlin/Java via Maven) and iOS (Swift via SPM). Built for mobile app developers and product teams that want to embed AI features directly on devices.
- Embedded AI Solutions Extended on-device AI deployment service that targets MCU through industrial computer hardware, enabling AI workloads to run on infrastructure at industrial sites without internet connectivity. Includes silicon adoption benchmarking for chipmakers, industrial infrastructure deployment, and hardware AI-readiness software integration services for organizations running AI at the edge.
Quantifiable outcome
- Up to 60x faster inference than CPU on NPU-accelerated devices
- +3 more outcomes
Companies that use ZETIC
Customer profileNamed customers25 records
Segments4 records
Ideal customer profiles4 records
ZETIC technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- No
- API docs
- API detail
Core technology
AI maturity
App detail
Integration4 records
AI capability8 records
Feature7 records
ZETIC partnerships and signals
Strategic signalScale indicators7 records
Recent moves5 records
Expansion highlights6 records
ZETIC competitors and assessment
Company assessmentDirect peers
- Qualcomm AI Hub: Qualcomm's own on-device AI model deployment and optimization platform aimed at Snapdragon devices. ZETIC overlaps most directly here — offering cross-vendor NPU deployment while competing head-on with Qualcomm's toolchain on its own silicon.
- Google AI Edge: Google's on-device AI deployment stack (including Gemini Nano, MediaPipe, TF Lite) for running models locally on Android. Closest direct peer to Melange's on-device deployment + SDK distribution model.
- MediaTek NeuroPilot: MediaTek's on-device AI SDK targeting Dimensity/MTK APUs. Direct overlap with Melange's MediaTek APU optimization layer; MediaTek has structural advantages on its own chipsets.
- OctoML (Apache TVM): Model compilation and deployment optimization across heterogeneous hardware (originally Apache TVM). Direct competitor to Melange's automated model optimization pipeline, with deeper open-source ecosystem traction.
- Deci (now part of Qualcomm): Deep learning optimization platform for inference efficiency, now absorbed into Qualcomm. Comparable positioning around automated quantization and acceleration, though Deci was acquired before scaling to Melange's mobile breadth.
- Fritz AI: Mobile machine learning platform for iOS/Android developers offering model training, conversion, and deployment tooling. Closest horizontal peer in mobile-developer-focused on-device ML infrastructure.
Broad incumbents
- Apple Core ML: Apple's platform-level on-device ML framework for iOS/macOS leveraging the Apple Neural Engine. ZETIC's iOS SDK competes against Core ML's deeper OS-level integration, making Apple the dominant incumbent on its own hardware.
- NVIDIA TensorRT: Industry-standard neural network inference optimizer. While primarily a server/desktop play, it spans edge/Jetson targets and is a comparable model-optimization incumbents benchmark against.
- PyTorch Mobile: PyTorch's official on-device inference runtime for iOS and Android. Operates as a backbone that ZETIC integrates with via the .pt2 export path, while also being an alternative inference path developers might pick instead.
Emerging players
- Hugging Face Optimum: Hugging Face's optimization toolkit for accelerating transformer models on-device (ONNX Runtime, TensorRT, TFLite). ZETIC's Hugging Face integration is effectively a channel partnership with a partially competing tool.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat6 records
Key risks6 records
Key highlights7 records
Customer concentration
ZETIC social profiles
Digital presenceZETIC compliance and trust
Trust signalCompliance3 records
ZETIC financial estimates
Financial estimateRevenue estimate
Valuation estimate
ZETIC leadership team
Management profileNumber of profiles
Profiles2 records
ZETIC funding detail
Funding detailFunding overview
Funding rounds4 records
Investors4 records
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
ZETIC M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about ZETIC
What does ZETIC do?
ZETIC provides the Melange platform (also marketed as MLange), an on-device AI deployment service that automatically optimizes, converts, quantizes, and compiles AI models (ONNX and PyTorch Exported Program) for hardware-accelerated execution on mobile CPUs, GPUs, and NPUs. The platform benchmarks models across 200+ real smartphone devices and delivers production-ready Android and iOS SDKs in under six hours via a three-step workflow (upload, benchmark, deploy), replacing months of manual NPU tuning.
Is ZETIC a public or private company?
ZETIC is a private company. It is classified as venture growth investor backed and is currently operating.
When was ZETIC founded?
ZETIC was founded in 2024. It employs 11 to 50 people.
Where is ZETIC based?
ZETIC is headquartered in San Francisco, United States, in the North America region.
How does ZETIC make money?
Five revenue lines are on record. SaaS Subscription - Free Tier is the primary driver. The others are saaS Subscription - Pro Tier, saaS Subscription - Pro+ Tier, saaS Subscription - Enterprise Tier and overage Fees.
Who are ZETIC's main competitors?
Direct peers on record are Qualcomm AI Hub, Google AI Edge, MediaTek NeuroPilot, OctoML (Apache TVM), Deci (now part of Qualcomm) and Fritz AI. Broad incumbents are Apple Core ML, NVIDIA TensorRT and PyTorch Mobile. Hugging Face Optimum is listed as an emerging player.
Does ZETIC have an API?
No public API is recorded for ZETIC.
What industry is ZETIC in?
ZETIC's product category is On-Device AI Deployment Platform. Its primary akta.pro industry code is HDAAABAF, Model Deployment, Serving & Inference Platforms, with a secondary code of HDAAAAAI, AI Compiler, Runtime & Kernel Optimization Software (CUDA/ROCm/XLA, graph compilers). Its NAICS code is 513210 and its SIC code is 7372.