TASMU Smart Qatar·Qatar National Vision 2030
الذكاء الاصطناعي السيادي لدولة قطر
The national marketplace for
Qatar's AI capability.
Discover models, datasets, prompts and agents published by Qatari entities. Evaluate before you commit, run on sovereign compute, and publish what you build, without a byte leaving the country.
0
Published assets
0
Entities
0.0M
Calls / 30 days
0%
In-country
2.5M inference calls served from sovereign infrastructure in the last 30 days.
What the platform offers
قدرة وطنية واحدة
Every AI capability the state owns, in one place
Thirty-one entities, five asset families, one set of rules. Instead of each ministry procuring its own models and rediscovering the same problems, capability is published once and reused across government.
Models
Arabic-first language, vision, speech and forecasting models trained on Qatari data and owned by Qatari entities.
2.5M calls served in 30 days
Datasets
Governed ministry data with published schemas, health scores and lineage, fitness for use judged in seconds.
Classification and refresh cadence on every record
Agents
Agents that execute real government workflows with declared tools, scoped data access and a named human at every consequential step.
20K runs in 30 days
Prompts
Reviewed, versioned prompt templates so thirty-one entities do not each rediscover how to ask the same question well.
Fork into your own namespace
Demos
Working mini-applications built on catalog assets, see a capability in a real service before you commit to it.
Runnable in the browser
Sandbox
Sealed compute to evaluate, compare and retrain against governed data, nothing leaves the sovereign boundary.
CPU included in your allocation
Trending on TASMU
What the country is actually using
Ranked by measured consumption rather than by whoever shouted loudest at procurement, with the credits each asset returned to the entity that published it.
Models
1fanar-gov-7b-instructRAG · Statistics520Kcalls / 30d2arabic-dialect-sentimentArabic NLP · Statistics418Kcalls / 30d3traffic-flow-predictorForecasting · Transport260Kcalls / 30d4khaliji-asrSpeech · Statistics210Kcalls / 30d5citizen-feedback-classifierArabic NLP · Statistics195Kcalls / 30dDatasets
1citizen-feedback-corpusJSONL · 412K rows8.4Kdownloads2air-quality-stationsCSV · 2.6M rows7.6Kdownloads3metro-ridership-dailyCSV · 94K rows6.9Kdownloads4establishment-census-aggregatesCSV · 84K rows5.2Kdownloads5population-projectionsCSV · 12K rows4.8KdownloadsAgents
1permit-review-agentDocument Processing · Identity5.1Kruns / 30d2citizen-triage-agentCitizen Service Triage · Statistics4.8Kruns / 30d3appointment-outreach-agentCitizen Service Triage · Healthcare2.7Kruns / 30d4incident-response-agentCitizen Service Triage · Transport1.5Kruns / 30d5census-data-qa-agentData QA · Statistics1.4Kruns / 30dPublish and earn
سرّع رحلتك في الذكاء الاصطناعي
Build it here, publish it nationally, earn it back
Any entity: a ministry team, an accelerator startup, a university lab, can build on sovereign compute and publish into the national catalog. What others consume comes back to you in credits.
Build in the sandbox
Sealed compute with governed datasets pre-mounted and the TASMU SDK installed. Nothing leaves the boundary.
Package and describe
Model card, intended use, limitations, lineage and licence, captured once, in the wizard, not in a document.
Clear the gates
Metadata, security scan, Responsible AI checklist and a named reviewer sign-off. Classification is inherited, not declared.
Published nationally
Discoverable by thirty-one entities the moment it clears. You keep ownership; they get reuse.
And then it earns
70% of every credit spent on your asset comes back to your entity
Consumption is metered in credits and settled against each entity's allocation, so no ministry has to invoice another. Publish capability that others use and you become a net creditor, carried forward, or converted into compute.
Sovereignty by construction
لا تعبر البيانات الحدود
The data never leaves Qatar
Sovereignty here is not a policy statement appended to a procurement document. It is how the platform is built: governed data, in-country compute, and inference all sit inside one boundary, and the paths that would cross it are closed.
- No egress from sandboxes
- Weights and data never leave in-country storage
- Derived models inherit the strictest upstream classification
- Every access purpose-bound and audited
Who publishes
Ministries, universities and startups, on the same shelf
An accelerator startup's model sits beside a ministry's, judged on the same published metrics and cleared through the same governance gates.
Open competition
قدرة تُكتسب بالمنافسة
The newest capability in the catalog was won, not bought
Unmet national needs are published as open challenges. Startups, universities and ministry teams compete on a governed benchmark, and the winning model is hardened through the Responsible AI pipeline and published for every entity to reuse.
Ramadan Traffic Flow Forecasting 2026
QAR 180,000 + pilot deployment with MOTC traffic operations
31 teams competing · scored on MAE (veh/15min)
Patient No-Show Prediction 2026
QAR 200,000 + fast-track to marketplace
38 teams competing · scored on AUROC
National Land-Use Mapping Challenge 2026
QAR 300,000 + fast-track to marketplace
Registration opening · scored on mIoU
From competition to catalog
2 past challenges have already delivered production models. Their winners are serving national programmes today: and earning credits for the teams that built them.
0
National challenges
0
Teams competing
QAR 0.00M
Awarded in prizes
Prefer the dense catalog view?
83 assets with filters, trending and the recently-published feed, the working screen for practitioners.