Fabryka.API

Meet the community

All events

    FABRYKA AI RESEARCH
    INDEPENDENT ARTIFICIAL INTELLIGENCE RESEARCH LABORATORY · WARSAW, POLAND

    We build models. Then put them to work.

    We study how language models learn, how to make them more efficient, and how to evaluate them reliably — with a particular focus on Polish-language AI.

    Find a use case →Explore Polish models →

    Illustration of a research studio with model training curves on a screen.
    MODELS · SYSTEMS · EVIDENCE

    NEW MODEL / REMIGIUSZ KINAS / BASAL 2.0 MAX

    Text in.
    One concrete decision.

    Basal 2.0 Max is an 11B decision model built on Bielik 11B v3.0. It reads up to 32,768 tokens of text and returns a calibrated decision. Try it live and connect it through the API.

    Explore Basal →

    Open weights · PL + EN · Stable release

    Illustration of the Basal model workshop

    FABRYKA AI USE CASE BOOK

    A Polish model.
    Your specific problem.

    Try working examples, explore API code and transparent measurements. Meet Basal, Muszka and Poziomka — each with clear authorship and actual availability.

    01 / CHOOSE A PROBLEM

    Try a use case

    Route support tickets, choose the next action and identify requests for a quote. Demos without an account, with results and code.

    Open the use case book →

    02 / MEET THE MODELS

    Polish models and creators

    A catalog of model families. Authors, sources, limitations and statuses: available, experimental and listed in the catalog.

    Browse the catalog →

    03 / ADD YOUR PROJECT

    Using a model? Show us how.

    Share an example, pilot or deployment. Describe the problem and the result — we will prepare a use case together.

    Fabryka AI is an independent service and is not organizationally affiliated with the creators of the featured models or with Hugging Face. Model names, brands and trademarks belong to their respective owners. Catalog inclusion does not imply sponsorship or endorsement of Fabryka by the authors. Model weights come from the authors’ Hugging Face repositories; hosting follows the license of the specific version. Each model card provides its source and license. A model license does not grant trademark rights. Model sources and licenses →

    FROM QUESTION TO PUBLIC ARTIFACT

    Research Experiments Evidence Applications

    01 / RESEARCH PROGRAMMES

    Three lines.
    One research agenda.

    How much quality can we achieve with a given budget of data, training and inference — and how can we measure it reliably? These three programmes connect our current work and planned studies.

    01 / RESEARCH PROGRAMME

    Small Language Models

    Training Polish language models from scratch. We study scaling laws, tokenizers, data mixtures and distillation to understand how much capability a compact model can learn.

    Sub-150M experiments · 1B / 10B research roadmap

    Explore programme →, Small Language Models

    02 / RESEARCH PROGRAMME

    Efficient Inference

    Making language models practical on commodity GPUs. We study quantization, speculative decoding, including DFlash, routing and the trade-offs between throughput, latency and quality.

    Quantization · decoding · hardware-aware evaluation

    Explore programme →, Efficient Inference

    03 / RESEARCH PROGRAMME

    Polish AI Evaluation & Data

    Building the data and measurements that Polish AI needs. This programme connects Polish DynaWord, CodeSOTA benchmarks, OCR evaluation and methodologies for comparing Polish LLMs.

    Datasets · benchmarks · reproducible measurement

    Explore programme →, Polish AI Evaluation & Data

    02 / SELECTED RESEARCH OUTPUTS

    Work you can inspect.

    Explore public model weights, data and benchmark evidence. Each linked artifact carries its own documentation and scope.

    MODEL

    Pollock Mini LM · 125M

    A compact language-model release. Model card, weights and usage information are available in the public repository.

    Model repository ↗

    DATASET

    Polish DynaWord

    A public Polish-language dataset supporting our work on language-model training and data composition.

    Dataset & documentation ↗

    BENCHMARK INFRASTRUCTURE

    CodeSOTA

    Model comparisons and benchmark evidence. Individual entries identify their sources and evaluation context.

    Explore benchmark evidence ↗

    Browse publications & research outputs →

    03 / OPEN RESEARCH

    Follow the work.
    Inspect the evidence.

    Our research connects questions, baselines, experiments and artifacts.

    Dated experiment write-ups are preserved in our research notes.

    Public experiment notes ↗

    04 / RESEARCH INFRASTRUCTURE

    Tools for doing
    the research.

    Training, measurement and serving are part of the same experimental workflow. We build the infrastructure that helps us investigate each step.

    MEASURE

    CodeSOTA

    Public infrastructure for model comparisons, benchmark results and evaluation evidence.

    Explore CodeSOTA ↗

    EXPERIMENT

    Compute laboratory

    GPU systems for training and inference experiments, with hardware and workload conditions documented alongside results.

    Inference research →

    05 / PEOPLE

    A lab is the people
    doing the work.

    Meet the Fabryka AI team: researchers, engineers and contributors working across models, data, evaluation, agents and infrastructure.

    Meet the Fabryka AI team ↓

    15
    team members & contributors
    2
    advisers
    Kacper Wikieł

    FOUNDER / RESEARCH

    Kacper Wikieł

    Language models, training infrastructure, model optimization and evaluation.

    CO-FOUNDER / RESEARCH & ENGINEERING

    Arkadiusz Słota

    Datasets, evaluations, training infrastructure and systems engineering.

    Piotr Zientara

    AI AGENTS / LLM INFRASTRUCTURE

    Piotr Zientara

    Founder of Xfaang. Builds autonomous AI agents and LLM products, connecting data, evaluation and infrastructure.

    Piotr Styła

    AI AGENTS / RECRUITMENT

    Piotr Styła

    Agent teams, web and mobile applications, recruitment and onboarding.

    Kuba

    AI ENGINEERING / FULLSTACK

    Kuba

    Theoretical physics, computer vision for oncology, and generative and agent systems.

    Kamil Kaczmarek

    FULLSTACK / MICROSERVICES

    Kamil Kaczmarek

    Bots, microservices and production tools, from idea to deployment.

    Maciej Sawicki

    EVALUATION / TOKENIZERS

    Maciej Sawicki

    Evaluation tooling, reproducible experiments and Polish byte-level BPE tokenizers.

    Dawid Majewski

    TRAINING INFRASTRUCTURE / PRODUCT

    Dawid Majewski

    Training infrastructure, developer tools and the Fabryka Track interface.

    Adam Skrodzki

    MODEL TRAINING / EVALUATION

    Adam Skrodzki

    Small-model training experiments, model-based evaluation and benchmark methodology.

    Paweł Puzio

    TOKENIZERS / TECHNICAL REVIEW

    Paweł Puzio

    Polish tokenizers, technical reviews and research planning.

    RESEARCH & PRACTICE

    Our advisers

    Supporting the team with research methodology, evaluation and domain expertise.

    06 / TECHNOLOGY TRANSFER

    Research,
    put to work.

    Selected technologies developed or evaluated at Fabryka become production systems. Our inference platform makes open models available through an API, with tools for chat and token-level analysis.

    The platform also serves models developed by other teams. Availability through our API is separate from authorship of a model.

    Explore Fabryka API →

    07 / ABOUT FABRYKA AI

    Independent research.
    Built in Poland.

    Fabryka AI is an independent artificial intelligence research laboratory in Warsaw. We build datasets, train models from scratch and develop methods for efficient inference and reliable evaluation.