Data residency for LLM APIs
Every model in the catalogue with the region its inference runs in and the retention its endpoint records, read from the endpoint rows the router itself filters on rather than from a marketing page. A model with no live endpoint is listed below as catalogue only, because that is what it is.
This deployment has no production inference backend. Every request is answered by the built-in mock backend in region dev-mock, so nothing is processed in the regions the endpoint rows record. The inference region is shown below as dev-mock for that reason, and no hosting class is claimed for a model that has no runtime here.
This page is documentation, not a certification. It records what each endpoint states about its region, its retention and its hosting class, and names the source. It is not a legal opinion, not a conformity assessment, and not a guarantee about any provider.
| Model | Provider | Inference region | Retention | Hosting class | Licence |
|---|---|---|---|---|---|
| Devstral Small 2 (24B) | Mistral AI | dev-mock | zero | no runtime here | apache-2.0 |
| EuroLLM-1.7B-Instruct | EuroLLM (utter-project) | dev-mock | zero | no runtime here | apache-2.0 |
| EuroLLM-22B-Instruct (2512) | EuroLLM (utter-project) | dev-mock | zero | no runtime here | apache-2.0 |
| EuroLLM-9B-Instruct (2512) | EuroLLM (utter-project) | dev-mock | zero | no runtime here | apache-2.0 |
| Mistral Small 4 (119B-A6B) | Mistral AI | dev-mock | zero | no runtime here | apache-2.0 |
| Whisper large-v3 | OpenAI | dev-mock | zero | no runtime here | apache-2.0 |
Catalogue only — no endpoint yet
These models are documented in the catalogue but have no endpoint that can answer a request, so there is no region and no retention to report for them. Asking for one returns an error rather than routing somewhere else.
- Aya 23 8B — Cohere
- Aya Expanse 8B — Cohere
- Codestral 25.08 — Mistral AI
- Command A+ (05-2026) — Cohere
- Command R (08-2024) — Cohere
- DeepSeek-R1 — DeepSeek
- DeepSeek-R1-Distill-Qwen-32B — DeepSeek
- DeepSeek-V3.2 — DeepSeek
- DeepSeek-V4-Flash (0731) — DeepSeek
- EmbeddingGemma 300M — Google DeepMind
- GLM-4.7-Flash — Z.ai (Zhipu AI)
- GLM-5.3 — Z.ai (Zhipu AI)
- GLM-OCR — Z.ai (Zhipu AI)
- Gemma 4 26B A4B IT — Google DeepMind
- Gemma 4 31B IT — Google DeepMind
- Gemma 4 E4B IT — Google DeepMind
- LLMEU V2 (announced) — LLM EU
- Llama 3.1 8B Instruct — Meta
- Llama 3.3 70B Instruct — Meta
- Llama 4 Maverick 17B-128E Instruct — Meta
- Llama 4 Scout 17B-16E Instruct — Meta
- Llama Nemotron Rerank 1B v2 — NVIDIA
- Magistral Small 1.2 — Mistral AI
- Ministral 3 8B — Mistral AI
- Mistral 7B Instruct v0.3 — Mistral AI
- Mistral Large 3 (675B) — Mistral AI
- Mistral NeMo 12B Instruct — Mistral AI
- NVIDIA Nemotron 3 Nano 4B — NVIDIA
- NVIDIA Nemotron 3 Super 120B-A12B — NVIDIA
- OpenEuroLLM final models (planned) — OpenEuroLLM
- OpenEuroLLM initial model release (planned) — OpenEuroLLM
- Pharia-1-Embedding-4608-control — Aleph Alpha
- Pharia-1-LLM-7B-control-aligned — Aleph Alpha
- Phi-4 — Microsoft
- Phi-4-multimodal-instruct — Microsoft
- Pixtral 12B — Mistral AI
- Qwen3-8B — Alibaba Qwen
- Qwen3-Coder-30B-A3B-Instruct — Alibaba Qwen
- Qwen3-VL-8B-Instruct — Alibaba Qwen
- Qwen3.8-27B — Alibaba Qwen
- Teuken 7B Instruct v0.6 — OpenGPT-X (Teuken)
- Voxtral Small 24B — Mistral AI
- Whisper large-v3-turbo — OpenAI
- gpt-oss-120b — OpenAI
- gpt-oss-20b — OpenAI
- gte-multilingual-reranker-base — Alibaba Qwen
- multilingual-e5-large — Microsoft