Bielik
Poland's open-source Polish-first LLM family, from 7B to 11B instruct models.
About
Bielik is the flagship Polish large language model family built by the SpeakLeash open-science community together with ACK Cyfronet AGH, trained on Polish supercomputers (Athena and Helios). The project started in April 2024 with Bielik-7B-v0.1, a Mistral-7B derivative continually trained on tens of billions of Polish tokens, and grew through an 11B v2 generation into v3 models with a dedicated Polish-optimized tokenizer.
The models are released under Apache 2.0 and rank at the top of the Open PL LLM Leaderboard for Polish linguistic and cultural understanding, reportedly outperforming much larger multilingual systems on Polish tasks. Quantized versions run on consumer GPUs as well as data-center accelerators, and the whole pipeline (data, training, evaluation) is developed in the open.
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 — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 109; Trending score 62/100 from researcher assessment.