Skip to content
HOSTACKER

AI

Coming soon

AI infrastructure without the complexity

Endpoints, inference and retrieval for developers. Private models and company knowledge for organisations. One key, one API, one invoice.

Explore Enterprise AIRead the docs

Developers

AI infrastructure for developers

Hosted endpoints, retrieval and model APIs behind one credential, so swapping a model is a configuration change rather than a rewrite.

  • AI Endpoints

    Coming soon

    A stable, OpenAI-compatible URL per project. Change the model behind it without touching your code.

  • Inference

    Coming soon

    Serverless serving for open-weight models, designed for per-token billing with no idle cost.

  • Embeddings

    Coming soon

    High-throughput embedding generation for search, clustering and recommendation pipelines.

  • Vector Databases

    Coming soon

    Managed vector collections on the same private network as your endpoints, so retrieval never leaves Hostacker.

  • AI APIs

    Coming soon

    Transcription, reranking, vision and moderation behind the same key and the same invoice.

AI products are not open for use yet. Pricing is published once inference capacity is contracted.

Enterprise

Private AI for enterprise

For organisations that want the capability without sending internal data to a shared, multi-tenant service.

Explore Enterprise AI

  • Private LLM

    Preview

    Isolated language-model infrastructure intended for company data and internal applications.

  • Enterprise RAG

    Preview

    Connect company documents, procedures and knowledge to a private AI experience with access control.

  • Dedicated Inference

    Preview

    Reserved or dedicated GPU infrastructure for AI workloads that need predictable throughput.

Also part of Enterprise AI

Enterprise knowledge indexing, MCP access for approved agents, and the security and governance controls that make both reasonable inside a company.

  • Enterprise Knowledge
  • Vector Infrastructure
  • MCP
  • SSO
  • Audit logs
  • Deployment policies

Platform

What the AI layer is built to give you

Not telemetry — Hostacker serves no traffic yet. These are the properties the platform is being designed around.

  • One key across every model

    Endpoints, embeddings and model APIs are designed to authenticate the same way, so swapping a model is a configuration change.

  • Usage visible per project

    Token consumption, request volume and cost are attributed per endpoint and per project rather than arriving as one opaque bill.

  • Retrieval on a private network

    Vector collections are planned to sit beside your endpoints, so documents and embeddings never traverse the public internet.

  • Cost estimated before you commit

    Every provisioning action is designed to return an estimate first, including the ones an AI agent initiates through MCP.

Models

Swap models without rewriting your integration

Endpoints are designed as stable URLs. Point one at a different model and your application keeps working.

ModelStatus
Llama 3.3 70B Instructhk-llama-3-70b
Preview
Llama 3.1 8B Instructhk-llama-3-8b
Preview
Mistral Smallhk-mistral-small
Coming soon
Qwen Coder 32Bhk-qwen-coder-32b
Coming soon
Hostacker Embed Largehk-embed-large
Coming soon
Intended model catalog. Availability and rates are confirmed when inference capacity is contracted.

Ship AI features, not AI plumbing

Tell us what you are building and we will map it to the AI infrastructure it needs.

Talk to HostackerExplore GPU