An open-source model whose weights can be self-hosted: Considering model-type differences, which option best
Only an open-source, self-hostable model provides offline operation with full weight inspection and no provider dependency.
The question
A defense contractor must run a language model entirely inside an air-gapped facility, retain full ability to inspect and modify weights, and avoid any dependency on an external provider's API. Considering model-type differences, which option best fits these constraints?
Preparing for AIGP? Take the free 5-min readiness quiz →
- A proprietary model accessed through the vendor's externally hosted API under a strong contractual data-processing agreementA hosted API requires external connectivity and offers no weight access, violating the air-gap and inspection requirements.
- An open-source model whose weights can be self-hosted, inspected and modified without any external provider dependency ✓Open weights allow fully offline hosting, inspection and modification with no vendor API dependency, satisfying every stated constraint.
- A proprietary multimodal model licensed for on-device use but with weights that remain closed to the licenseeClosed weights, even if deployable locally, block the required inspection and modification of the model.
- A small classic machine-learning classifier retrained on internal documents for narrow prediction tasksA classic classifier is not the language model the use case needs, so it fails the functional requirement outright.
The trap
Assuming an on-device or locally licensed proprietary model grants the same control as open weights when its weights stay closed. How to remember it
Only an open-source, self-hostable model provides offline operation with full weight inspection and no provider dependency.
How many of these would you get right?
One of 1581 AIGP questions on Certsqill. Take a free five-minute check and see your score per domain — not one number, but which section to open tonight.
Test your AIGP readiness — freeMore Understanding How to Govern AI Deployment and Use questions
- Workforce readiness: Before approving deployment, which contextual factor is the most critical gap to resolve? →
- Retrieval augmented generation that grounds responses: Which deployment technique best satisfies these →
- Performing an impact assessment that evaluates effects: Which activity is being described? →
- All 424 Understanding How to Govern AI Deployment and Use questions →
Part of the Certsqill AIGP question bank · Understanding How to Govern AI Deployment and Use ·
Every answer, right and wrong, comes with its own explanation.
