Bielik 4.5B v3
Efficient Polish mid-size model from the Bielik v3 generation.
About
Bielik 4.5B v3 is the compact tier of the Bielik v3 model generation, developed by the SpeakLeash project with ACK Cyfronet AGH. At 4.8 billion parameters, it extends the v3 tokenizer and training improvements down to a size that fits comfortably on consumer hardware.
In the Bielik v3 evaluation suite it scored competitively for its size class on Polish benchmarks, offering a practical middle ground between the 11B flagship and the 7B Minitron distillation. Like the rest of the v3 line it is trained with a heavy emphasis on curated Polish corpora plus English data to prevent catastrophic forgetting.
Released under Apache 2.0, it is aimed at developers who need quality Polish language performance without the infrastructure cost of the 11B model: local chatbots, document processing, and Polish-language RAG pipelines.
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 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 99; Trending score 70/100 from researcher assessment.