Quick answer. Mistral Large 3 has a weighted score of 83, privacy 92 and EU jurisdiction. It is self-hostable. GDPR, HIPAA and SOC 2 are true. The API rate is $2 in and $6 out per million tokens. Choose it when data sovereignty or an on-prem path outranks US frontier brands.
Mistral Large 3 is the European balanced row in the published Best AI Match dataset, ranked 11 of 21 with a weighted score of 83. This profile decides whether EU jurisdiction and self-hostability justify picking Mistral over cheaper or higher-task US APIs. It is not a claim that every EU workload must use Mistral, and it does not invent a private sovereignty bench. Privacy is 92. Task performance is 83. Truth Score is 80. Cost efficiency is 82 at $2 per million input tokens and $6 per million output tokens. Context is a 128,000-token window, scored 78. Speed is 84. The model is self-hostable and jurisdiction is the European Union. GDPR, HIPAA and SOC 2 are marked true. Perspective Score is minus 5 and stays outside the weighted score. Best-fit jobs in the dataset are EU data sovereignty, regulated industries, government and defence, and self-hosted open weights. Those jobs are the gate. If the brief only needs the lowest token price without a sovereignty constraint, the scorecard's cost lens may point elsewhere.
Sovereignty is why this profile exists
Privacy 92, EU jurisdiction and self-hostable true are the decision lenses, not the weighted rank of 11. Best-fit jobs name EU data sovereignty, regulated industries, government and defence, and self-hosted open weights. Read the landscape on Europe AI landscape and AI data sovereignty comparison before you treat a US API as interchangeable.
| Sovereignty and privacy lens | Published figure |
|---|---|
| Weighted score (rank 11 of 21) | 83 |
| Privacy | 92 |
| Task performance | 83 |
| Truth Score | 80 |
| Cost efficiency | 82 |
| API price, input / output per million | $2 / $6 |
| Self-hostable | true |
| Jurisdiction | European Union |
| GDPR / HIPAA / SOC 2 | true / true / true |
| Perspective Score (excluded) | −5 |
Figures from /data/ai-comparison-2026.json. Last verified 2026-06-26. Scoring version v1.0. Affiliate links are off.
Cost versus DeepSeek's sticker
Cost efficiency is 82 at $2 and $6 per million tokens — higher than DeepSeek V3's $0.27 and $1.10, and far below Claude Opus 4.8's $5 and $25. The enterprise question is often privacy spend, not the absolute cheapest token: how much enterprise AI privacy costs and when a cheaper model is safer for customer data.
Self-host path
Self-hostable true means the privacy score can be paired with infrastructure you operate. That is the opposite hosted story from DeepSeek V3's China-jurisdiction API with privacy 52. Compare the deployment decision on self-hosted vs API AI and keep day-to-day controls on the privacy checklist.
Official page, affiliates off
Perspective Score is −5. It does not move the weighted score. The official site is mistral.ai, linked as the provider page while affiliates stay off.
Match under a sovereignty constraint. Run the match engine with privacy weighted for the job. Compare DeepSeek V3 only when self-hosting is real, or return to the model index.
FAQ
What is Mistral Large 3's weighted score?
Mistral Large 3 has a weighted score of 83 and is ranked 11 in the published Best AI Match dataset, last verified 2026-06-26.
Is Mistral Large 3 self-hostable?
Yes. The published record marks Mistral Large 3 as self-hostable, with European Union jurisdiction.
What does the Mistral Large 3 API cost?
The dataset lists $2 per million input tokens and $6 per million output tokens.