Request
Add T-Systems AI Foundation Services (LLMHub) as a new provider in models.dev.
Why
T-Systems LLMHub is a GDPR-compliant, EU-hosted LLM gateway run by Deutsche Telekom's enterprise IT subsidiary. It exposes 38 models — a mix of frontier proprietary models (OpenAI, Anthropic, Google, Mistral) and open-weight models hosted on T-Cloud Germany (Llama, Qwen, GPT-OSS, Teuken). It is widely used by European enterprises that need EU data residency for AI workloads, and currently has no entry in models.dev.
Provider details
Proposed model coverage (38 total)
| Category |
Count |
| LLMs |
30 |
| Embeddings |
5 |
| Audio (Whisper) |
2 |
| Image generation |
1 |
Status
Implementation submitted in PR #1383. The PR uses `extends` (README rule 3a) for 26 of 38 wrapper models, inheriting from canonical definitions in `providers/{anthropic,openai,google,mistral,llama,alibaba}/` with T-Systems pricing markup as local overrides. Validation passes (`bun run validate`) and `bun run compare:migrations` shows zero diff for the conversions.
Request
Add T-Systems AI Foundation Services (LLMHub) as a new provider in models.dev.
Why
T-Systems LLMHub is a GDPR-compliant, EU-hosted LLM gateway run by Deutsche Telekom's enterprise IT subsidiary. It exposes 38 models — a mix of frontier proprietary models (OpenAI, Anthropic, Google, Mistral) and open-weight models hosted on T-Cloud Germany (Llama, Qwen, GPT-OSS, Teuken). It is widely used by European enterprises that need EU data residency for AI workloads, and currently has no entry in models.dev.
Provider details
https://llm-server.llmhub.t-systems.net/v2Proposed model coverage (38 total)
Status
Implementation submitted in PR #1383. The PR uses `extends` (README rule 3a) for 26 of 38 wrapper models, inheriting from canonical definitions in `providers/{anthropic,openai,google,mistral,llama,alibaba}/` with T-Systems pricing markup as local overrides. Validation passes (`bun run validate`) and `bun run compare:migrations` shows zero diff for the conversions.