SCX-coder
From Code Generation to Software Execution
An Australian sovereign coding model built for agentic software development, high-throughput inference, and ultra-low-cost deployment.
SCX-coder is a high-performance coding model available now on the SCX API. Designed for agentic coding workflows, it can plan, execute, test, and refine software development tasks across real engineering environments.
The Challenge with Current Coding Models
Focused on autocomplete and isolated code snippets
Often require expensive proprietary infrastructure
Limited ability to execute multi-step development workflows
High latency and cost make long-running agentic tasks difficult
May route sensitive code or development context offshore
Built for agentic software development workflows
Can break complex engineering tasks into structured steps
Designed to generate, test, and refine code iteratively
Runs on SCX sovereign AI infrastructure in Australia
Available now through the SCX API
“We’re moving beyond copilots into fully agentic systems that can execute software development workflows. SCX-coder is built to operate inside real engineering environments — not just assist them.”
- David Keane, CEO & Co-Founder, SCX
Our Technical Innovation
Built for Agentic Software Development
Unlike conventional coding assistants, SCX-coder is designed for software execution: the ability to plan, act, iterate, and maintain context across real development workflows.
Breaks complex engineering tasks into structured steps before generating code.
Supports long-running development workflows across files, APIs, tools, and environments.
Generates, tests, and refines code through repeated development loops.
Maintains context across codebases, systems, APIs, and software pipelines.
Adapts tool usage across development environments and agent frameworks.
Runs on SCX sovereign infrastructure with no offshore data routing or retention.
Performance Built for Real Engineering Workloads
Up to 6×
SCX-coder is designed to deliver up to 6× higher throughput than leading proprietary models, enabling faster agentic coding workflows.
~41%
Lower latency helps agentic workflows move faster through planning, execution, testing, and refinement loops.
80–90%
SCX-coder is designed to reduce the cost of software development workloads, especially for high-volume coding and agent tasks.
Up to 10×
Powered by efficient inference infrastructure, SCX-coder supports a more sustainable cost and energy profile for enterprise AI development.
Built on Project MAGPiE: Australia’s AI Framework
SCX-coder joins Project MAGPiE, SCX’s broader initiative to develop Australian-aligned AI systems optimised for local enterprise, government, and developer environments.
Defaults to Australian English conventions in code, documentation, and outputs
Aligns with local infrastructure, governance, and regulatory expectations
Avoids US-centric assumptions in system design and deployment patterns
Reflects the workflows and standards of Australian engineering teams
Powered by SCX Sovereign Infrastructure
SCX-coder is deployed on SCX’s sovereign AI infrastructure, leveraging ASIC-accelerated inference powered by SambaNova systems. It is designed for secure, low-latency, high-throughput deployment without routing data offshore.
Onshore deployment in Australia
No offshore data routing
No data retention
No training on customer data
ASIC-accelerated inference
Private interconnection support via Equinix Fabric
Suitable for enterprise and government codebases
Built for Modern Software Teams
Technical Specifications
Start Building with SCX-coder Today
SCX-coder is available now on the SCX API. Add sovereign, high-performance coding intelligence to your developer tools, agent frameworks, and enterprise software pipelines.
Available Now
OpenAI & Anthropic Compatible API
Sovereign Australian Infrastructure