Go EuropeAIdirectory

Evidence-backed European AI register

kaiko.ai

Reviewed identity, roles, model listings, pricing and evidence for kaiko.ai.

Dataset compiled 2026-08-27 · evidence over inference

model lab3 model listings0 token-price rows

Reviewed identity

Headquarters scope
European non-EU headquarters
Headquarters record
Zurich (with a significant Amsterdam presence)
Legal entity
AG
Ownership
privately funded; backed by Nadia Schadlow-linked and Dutch/Swiss investors, with a long-running partnership with the Netherlands Cancer Institute
Founded
2020
Confidence
medium
Last verified
2026-08-22

Links

Official website

Open interactive profile

Evidence boundaries

Organisation listings identify a reviewed canonical record. Model listings show an association in the source register; they do not by themselves establish model creation, current API availability, hosting location or infrastructure ownership.

Model listings

  1. Midnight-12kcatalogued with this organisation · origin not asserted
    Type
    multimodal
    Parameters
    unknown
    Context
    unknown
    Languages
    n/a — histopathology whole-slide images
    Licence
    MIT
    Release
    2025-04
  2. kaiko pathology foundation models (ViT-S/B/L, DINO/DINOv2-based, TCGA-trained)catalogued with this organisation · origin not asserted
    Type
    multimodal
    Parameters
    ViT-S to ViT-L class
    Context
    unknown
    Languages
    n/a — histopathology
    Licence
    open source
    Release
    2024-04
  3. eva (open-source evaluation framework for pathology foundation models)catalogued with this organisation · origin not asserted
    Type
    text
    Parameters
    unknown
    Context
    unknown
    Languages
    n/a — benchmark harness
    Licence
    Apache-2.0
    Release
    2024

Profile note

Zurich medical AI lab that releases genuinely open (MIT) pathology foundation models — Midnight-12k matches or beats Virchow2 and UNI-2 while training on orders of magnitude less data — and maintains eva, the open evaluation framework the whole field now benchmarks against. Directly comparable to Berlin's Aignostics but far more open. No pricing; open weights.

Record sources

  1. https://kaiko.ai/research/
  2. https://github.com/kaiko-ai/midnight
  3. https://arxiv.org/abs/2404.15217
  4. https://arxiv.org/html/2504.05186v1
  5. https://kaiko.ai/insights-hub/towards-training-large-scale-fms/
  6. https://kaiko.ai/