Sandbox & AI Toolkits
Sealed experimentation environments on sovereign compute, evaluate, compare, and retrain without data leaving the platform boundary.
Ministry of Public Health, monthly sandbox quota
68% of 200 compute-hours used · resets in 2 daysHealthy
Most used
Model test-bench
Evaluate any catalog model against a governed dataset or your own CSV, inside a sealed sandbox.
Side-by-side comparison
Run one input through two models and compare outputs, confidence, and latency in parallel.
New
No-code retrain
AutoTrain-style wizard: pick a base model, point at fresh data, launch a governed retraining run.
Notebook environment
Managed JupyterLab with pre-mounted governed datasets and the TASMU SDK.
Compute environments
| Environment | Specification | Status | Indicative price |
|---|---|---|---|
| CPU · Standard | 8 vCPU · 32 GB | Available | Included in quota |
| GPU · Shared T4 | 1× T4 · 16 GB VRAM | Available | QAR 4.2 / hour |
| GPU · A100 Slice | 1× A100 40 GB (MIG) | 2 in queue | QAR 28 / hour |
| GPU · A100 Dedicated | 8× A100 80 GB node | By reservation | QAR 190 / hour |
All sandboxes are network-isolated: no internet egress, governed datasets mounted read-only, outputs scanned before export. Sessions auto-terminate after 8 idle hours.