Hosn Kernel
A single-appliance starting scope for a focused workload. The final model, capacity, integrations and delivery terms follow measured requirements.
What Hosn Kernel is
Hosn Kernel is the single-appliance Hosn configuration for a department-level on-premise AI workload. It is installed at the customer's site. The final hardware, model, storage, capacity and integrations are selected only after the workload and operating boundary have been reviewed.
Kernel is a configuration name, not a promise of one fixed bill of materials. The written delivery scope is the authority for what is supplied, how it is connected, and how acceptance is tested.
Public configuration summary
| Configuration | Single-appliance deployment |
|---|---|
| Intended scope | Department-level workloads |
| Model | Selected and validated during discovery |
| Data boundary | Customer site, with an air-gapped option |
| Capacity | Sized against the measured workload |
| Integration | Defined in the delivery scope |
| Commercial terms | Written quotation and contract |
When Kernel may fit
Kernel is intended for a focused workload with a limited user group and a defined corpus. Typical examples include internal document search, Arabic drafting support, summarisation and retrieval-assisted questions over approved material. These are workload patterns, not customer references or guaranteed capabilities.
If the measured concurrency, corpus size, availability requirement or integration scope exceeds a single-appliance design, Hosn Tower or Hosn Rack may be a better starting point.
How the model is selected
Hosn can evaluate selected open model families, including Gemma and Qwen, against a representative and approved sample. The exact variant, runtime and context setting are recorded in the delivery scope. No model is described as suitable until its output, latency, memory use and language behaviour have been tested for the workload.
Data and network boundary
The deployment can be designed for an isolated local network, including an air-gapped option. Required directory, logging, storage and user-network connections are mapped during discovery. The implemented boundary and update process are then documented for the institution's technical and security review.
What a quotation confirms
A written quotation confirms the bill of materials, software scope, implementation, integrations, acceptance criteria, schedule, support and commercial terms. Hosn does not publish a public price because the scope is configuration-specific.
Read next
- Mac Studio M3 Ultra LLM Inference Guide
- Apple Silicon LLM Benchmarks: M3 Ultra vs M4 Max
- Sizing a Sovereign AI Appliance: Concurrent Users, Latency, and Throughput
- Gemma 4 with 256K Context: A Deep Dive for Sovereign On-Premise Deployment
- Qwen 3.6 Arabic NLP Benchmarks, Qwen 3.8 Update
- Edge GPU Appliances for Branch and Field Offices
Next step. Start with a scope discussion involving the institution's IT and data owners. The readiness assessment documents the workload, data boundary, integrations and delivery requirements before a written quotation is prepared.