Teuken-7B
European 7B LLM trained from scratch in 24 EU languages.
About
Teuken-7B is the OpenGPT-X consortium's 7B-parameter language model, trained from scratch on 4T tokens with roughly half non-English data across all 24 official EU languages. Led by Fraunhofer IAIS and IIS with partners including Aleph Alpha and IONOS, it was built as a European-values alternative to US- and China-centric models, and the commercial variant ships under Apache 2.0.
Its strength is multilingual European coverage, not raw capability: the team itself lists math and coding as out of scope, and it trails larger multilingual models on hard benchmarks. Note the licensing split — only the commercial variant is Apache 2.0; research and non-commercial variants have their own terms.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (verified tier, metrics-v1). Evidence: OpenGPT-X consortium led by Fraunhofer, Nov 2024; trained from scratch on 4T tokens across 24 EU languages; Commercial variant is Apache 2.0; research and non-commercial variants have separate terms — pick the right one; Team is explicit about scope: math and coding are out of scope, and it trails larger multilingual models on hard benchmarks; Niche but real value for EU-language coverage where US/China-centric models underperform. License: Apache-2.0 applies to the -commercial variant only; other variants have their own terms.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 6.0/10
- Maintenance · recency, releases, issue load
- 6.0/10
- Openness · license + docs
- 7.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · OpenGPT-X consortium led by Fraunhofer, Nov 2024; trained from scratch on 4T tokens across 24 EU languages
- · Commercial variant is Apache 2.0; research and non-commercial variants have separate terms — pick the right one
- · Team is explicit about scope: math and coding are out of scope, and it trails larger multilingual models on hard benchmarks
- · Niche but real value for EU-language coverage where US/China-centric models underperform