OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Infrastructure & Serving

Atlas — Avarok-Cybersecurity

Atlas is an open-source inference engine written in Rust and CUDA that runs large language models and serves requests through OpenAI-compatible, Anthropic and Responses APIs. Its capabilities include hardware-specific CUDA kernels, NVFP4 and FP8 quantization, speculative decoding, radix-tree prefix caching and expert parallelism across nodes.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Atlas — Avarok-Cybersecurity — OpenViST project profile in AI Infrastructure & Serving.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Atlas when you need to operate LLM inference yourself and want API-compatible serving with an engine whose documented stack includes Rust, CUDA kernels and model-specific optimization features.

Who it is for

  • Developers building applications that need an OpenAI-compatible, Anthropic or Responses inference endpoint.
  • Teams operating self-hosted LLM inference, including agentic coding workflows.
  • Engineers tuning inference deployments for particular hardware, models and quantization choices.

Documented uses

  • Run a supported language model through Atlas's Rust inference engine.
  • Serve model inference to clients using compatible API formats.
  • Choose a documented command-line or Docker setup path.
  • Use quantization, speculative decoding or prefix caching as part of an inference configuration.
Before you choose

Requirements & limits

The project is open source under the AGPL-3.0 license. The repository documents a prebuilt atlasctl installer and a Docker image for GB10.

  • Hardware and model matter: kernels are tuned for hardware, model and quantization combinations.
  • The README describes verification on NVIDIA DGX Spark with GB10/Blackwell and compilation for AMD Strix Halo through SCALE.
  • The interactive Docker Library requires a real terminal; the example also passes through GPU access and uses host networking and shared memory.
  • The installer places atlasctl in ~/.local/bin and does not require a Rust toolchain; building via Cargo is also documented.
  • AGPL-3.0 is the stated project license.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install the Atlas command-line tool

Install the prebuilt atlasctl binary using the documented shell installer.

Before you start

  • A shell environment that can run the installer and download the prebuilt binary.

Steps

  1. Run the command below.

    Documented command
    curl -fsSL https://atlascybernetics.ai/install.sh | sh

  2. The installer downloads atlasctl, verifies its checksum and installs it in ~/.local/bin.

Expected outcome

The atlasctl command-line tool is installed in `~/.local/bin`.

Guide limitations

  • This procedure documents installation only; it does not establish that a particular model will run on the host.
  • The documented alternative is cargo install atlasctl, which builds from source.
Documentation 1 ↗
Source-based practical guide

Start the Docker model library on a GB10

Launch Atlas's interactive model library using the documented GB10 Docker configuration.

Before you start

  • A GB10 GPU system with Docker GPU access.
  • A real terminal for the interactive text interface.
  • Docker shared-memory and host-network options, plus the host cache and Atlas-state directories.

Steps

  1. Run the README's the command below command with avarok/atlas-gb10:latest serve, --gpus all, --ipc=host and --network host.

    Documented command
    docker run

  2. Mount ~/.cache/huggingface at /root/.cache/huggingface and ~/.avarok at /root/.avarok as shown in the command.

  3. In the terminal interface, choose a model and recipe from the Library.

Expected outcome

The container starts the interactive Library so a model and recipe can be selected.

Guide limitations

  • The documented interactive selection requires a TTY.
  • This recipe is for the README's GB10 image and does not establish support for other hardware.
Documentation 1 ↗
COMPARE OPTIONS

Compare options for Model inference and serving

These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.

ProjectDocumented purposePlatformAccess and requirements
Atlas — Avarok-CybersecurityRun a supported language model through Atlas's Rust inference engine.Cloud, Linux, Self-hosted, ServerThe project is open source under the AGPL-3.0 license. The repository documents a prebuilt atlasctl installer and a Docker image for GB10.
MorphingdbImport a libtorch model and create predictions from values in a PostgreSQL table.Docker, Linux, Self-hosted, ServerThe README documents a Docker image and a build-from-repository option, followed by connecting to PostgreSQL and enabling the pgdl extension.
Gateway — pyadavSend chat-completion requests to a provider through a common server endpoint.ServerThe repository is public and identifies the project as Apache-2.0 licensed. README installation instructions include binaries, platform package managers and Docker.
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
13 October 2026
Original website title
Avarok-Cybersecurity/atlas · Official homepage title · retained in its source language
How we review sourcesSuggest a correction