GitHub hosts HunyuanVideo, Tencent's open-source framework for large-scale video generation models, enabling AI-driven video creation.
Acrux Core
FreeStarting price
0Popularity
About Acrux Core
Acrux Core is an open-source LLM-ops platform built for engineering teams working with large language models. Positioned between applications and model providers, it provides a unified control plane for managing prompt versions, routing traffic via an OpenAI-compatible gateway, and capturing distributed traces of every AI call. The platform automatically logs detailed spans for each request, including model used, token counts, latency, and cost, enabling teams to analyze performance and expenses without manual instrumentation. Acrux Core supports production workflows by allowing teams to update prompts and shift traffic between model versions using production aliases, eliminating the need for application redeployments. It offers TypeScript and Python SDKs, a REST API, and full self-hosting capabilities under the Elastic License 2.0, making it suitable for teams prioritizing data control and customization. Engineering teams use Acrux Core to standardize prompt management, reduce debugging time, and maintain audit trails for AI-driven features across development and production environments.
Key features
- Prompt versioning with production alias management
- OpenAI-compatible gateway for seamless integration
- Distributed tracing of every LLM call with span-level details
- Automatic logging of model, tokens, latency, and cost
- TypeScript and Python SDKs for developer tooling
- REST API for programmatic control
- Full self-hosting under Elastic License 2.0
- Evaluation pipelines for testing prompts and models
- Tool catalogs for managing reusable AI functions
- Zero-downtime updates to prompts and model routing
Use cases
- Versioning and deploying prompts across development and production
- Debugging and optimizing AI call performance and costs
- Enforcing governance and audit trails for AI-driven applications
Pros
- Unified control plane for managing prompts, gateway routing, tracing, and evaluation in one platform
- Open-source under Apache License 2.0, enabling self-hosting and full transparency
- OpenAI-compatible gateway allows drop-in replacement without modifying existing client code
- Automatic distributed tracing captures model, tokens, latency, and cost for every call
- Supports production workflows with versioned prompts and aliases to update traffic without redeploying applications
Cons
- Currently in beta, which may imply limited stability or support for production workloads
- Self-hosting requires infrastructure management and maintenance effort
- Limited to teams comfortable with integrating SDKs or modifying API endpoints
Frequently asked questions about Acrux Core
What does Acrux Core do?
Acrux Core is an open-source LLM-ops platform that sits between applications and model providers, offering versioned prompt management, an OpenAI-compatible gateway, distributed tracing, and evaluation tools for engineering teams.
Who is Acrux Core suitable for?
It is designed for engineering teams working with large language models who need standardized prompt management, debugging, and audit trails across development and production environments.
How does Acrux Core handle pricing and model costs?
Users bring their own provider keys, so model spend remains theirs, and Acrux Core does not take a cut of tokens. The platform itself is free during the beta phase.
Can Acrux Core integrate with existing applications?
Yes, it provides an OpenAI-compatible gateway, allowing existing OpenAI clients to route calls through Acrux Core without code changes.
Does Acrux Core support self-hosting?
Yes, the entire platform can be self-hosted under the Apache License 2.0, giving teams full control over their data and infrastructure.
What SDKs and APIs are available?
Acrux Core offers TypeScript and Python SDKs, as well as a REST API, enabling integration with various programming environments.
