Fixstars Launches Secure AI Appliance Fixstars Vega for Embedded Development Companies

Fixstars Corporation (TSE Prime: 3687, CEO: Satoshi Miki), a leading company in performance engineering technology, is launching the AI-driven development appliance "Fixstars Vega," which offers high security and low operational costs for companies engaged in embedded software development.

Fixstars Vega is an appliance that allows verified and optimized state-of-the-art open-weight large language models (open-weight LLMs) [1] to be operated securely in an on-premises or private cloud environment dedicated to the customer. Furthermore, it supports the utilization of AI in embedded product development by incorporating embedded software development expertise cultivated by Fixstars over 20 years as a performance engineering harness [2].

Fixstars Vega Overall Configuration

Features of the Secure AI Appliance "Fixstars Vega" for Embedded Development Companies

Embedded development sites, which develop for systems with strict security requirements and high levels of specialization and uniqueness, face the challenge of difficulty in using general cloud-based AI services.

Fixstars Vega is an appliance product that addresses these challenges with advanced security and technical capabilities.

Feature 1: Unlimited AI Usage Without Token Charges

Since open-weight LLMs are operated in the customer's dedicated local environment (on-premises, private cloud, etc.), there are no token-based usage fees like those associated with cloud AI services. Fixstars will continuously verify new models as they become available and provide them as an appliance that maximizes hardware performance.

Feature 2: High Security Achieved by Building AI in a Dedicated Corporate Environment

AI environments can be constructed according to the confidentiality of the information handled and internal policies, such as on-premises, hosted private cloud, or cloud (VPC type), allowing AI to be utilized without exposing confidential information such as source code and design information to external networks. There is also no risk of input data being learned by external AI services.

Feature 3: Unique Harness Supporting Efficiency and Quality Improvement in Embedded Development

A harness is a set of software functions, such as context management and peripheral tool operation, necessary for an AI model to perform processing appropriately and efficiently. It forms an AI agent in conjunction with the LLM.

Fixstars Vega incorporates the performance engineering know-how accumulated by Fixstars over 20 years as a harness.

By using AI agents that combine this unique harness with verified state-of-the-art open-weight LLMs, it is possible to generate code that achieves performance optimization, which is not possible with general AI.

Speed-up Examples with Fixstars Vega

Image Segmentation Foundation Model Inference: 4.7x speed-up

SparseDrive Inference: 4.0x speed-up

ResNet-50 Training: 2.1x speed-up

Fixstars Vega Configuration Overview

Fixstars Vega is an appliance that builds and provides a dedicated AI development environment for customers on-premises or in a private cloud.

Additionally, we offer detailed consultation, implementation support, and training services to propose and implement the optimal AI development environment for each customer.

Main Components and Functions

AI Infrastructure: Based on your requirements, we will prepare the optimal environment for you, whether on-premises or in a private cloud.

Open-Weight LLM: Rather than simply providing available LLMs, we provide LLMs optimized for embedded developers after verification and optimization. Fixstars Vega Standard LLMs (as of July 2026): GLM-5.2, Qwen3.6-27B

Gateway for Secure AI Environment: This gateway temporarily receives AI inquiries from each user and routes them to safe AI models based on the customer's pre-configured security policies.

Performance Engineering Harness: This is Fixstars' proprietary harness that optimizes AI output for embedded development, including profiling, bottleneck identification, operation optimization, and device-specific optimization.

AI Coding Environment: This includes peripheral software such as coding assistants and chat UIs necessary for AI-driven development. It is composed of open-source applications.

Ancillary Services

Design and Construction of Secure AI Infrastructure: We design and build environments tailored to security requirements, such as on-premises, hosted private cloud, or cloud (VPC type).

Introduction of AI Coding Environment: We support the introduction and utilization of tools beneficial for development tasks, including AI chat, AI coding agents, and IDE integration.

Selection, Evaluation, and Operational Design of AI Models: AI models are constantly evolving. The Fixstars Vega team will confirm operation and verify future releases, providing state-of-the-art models optimized for hardware.

Design for Utilization of Internal Knowledge: We create mechanisms for AI to reference specifications, design documents, internal rules, and past materials.

Training and Operational Adoption Support: We support AI utilization through developer seminars, establishment of usage rules, and visualization of usage status.

Fixstars Vega Details: https://www.fixstars.com/ja/products-services/vega

[1] Open-Weight LLM

An open-weight LLM is a large language model whose trained weights (parameters) are publicly released, allowing anyone to download and run it in their local environment. In recent years, its use as a "local LLM," running models in dedicated environments such as personal computers, on-premises, or private clouds, has gained attention.

[2] Harness

A harness is a set of software functions, such as context management and peripheral tool operation, necessary for an AI model to perform processing appropriately and efficiently. It forms an AI agent in conjunction with the LLM.

About Fixstars Corporation

Fixstars is a technology company with the corporate message "Speed up your AI." Leveraging software optimization technology that maximizes computational resources, we achieve overwhelming speed-ups in both AI model inference processing and learning processes. We are driving the evolution of next-generation AI technologies in various fields, including healthcare, manufacturing, finance, and mobility.

https://www.fixstars.com/