AI/LLM
Proximity
Native (own / self-hosted weights)
Models
proprietary / self-built models
Cloud · datastore
Multi-Cloud
Compliance
—
Architecture
AI21 Labs develops and serves its own Jamba family of LLMs and related services via its AI21 Studio API and private deployments, while also distributing Jamba models through third‑party clouds such as Azure and upcoming NVIDIA APIs.[5][6][7] Public materials describe deployment options across public cloud and private/on‑prem environments but do not reveal a full production stack beyond this high‑level multi‑cloud posture.[3][5][6]
AI/LLM
Proximity
Native (own / self-hosted weights)
Models
Mistralproprietary / self-built models
Cloud · datastore
Multi-Cloudobject storage for model artifacts, datasets, and logs (e.g., S3-compatible across partner clouds)[10]vector databases used with Mistral embeddings and open-weight models (e.g., Qdrant, Pinecone, pgvector; typical for La Plateforme and self-hosting scenarios)[11]relational databases for accounts, billing, and orchestration metadata (likely PostgreSQL/MySQL, inferred from typical SaaS/API patterns)[10][11]
Compliance
SOC 2 Type II (claims to comply with SOC 2 Type II framework, with reports available via Trust/Help Center)[3][4][7][12][15]ISO 27001 (claims to comply with ISO 27001 framework)[3][4][7][12][15]ISO 27701 (claims to comply with ISO 27701 framework)[3][4][7][12][15]GDPR (EU-incorporated company with GDPR-focused data processing and SCCs; EU data residency by default for La Plateforme)[3][4][6][9][11]
Architecture
Mistral runs a multi-cloud architecture where its hosted La Plateforme and Studio offerings sit atop partner clouds (Google Cloud, AWS, Azure, SAP, IBM, Snowflake, NVIDIA, Outscale), while open-weight models are also deployable on‑prem via standard inference containers.[10][11] The production stack is therefore split between first‑party EU‑centric hosting for regulated workloads and distribution of models through major cloud marketplaces and managed services.