European Synthetic Data Platforms

#1 Octopize

Nantes, France Free for one month; Starter €15k, Professional €48k, Expert €83k (billing period not stated on the page) Demo on request

Best for: Health, industry and defence organisations that must fix a budget before the first sales call

• Operating company. Octopize Mimethik Data (SAS)
• Jurisdiction. EU (France)
• Where the data sits. Scaleway cloud in France (HDS certified per vendor) or your own infrastructure
• Source code. Closed source
• Replaces. Gretel, NVIDIA NeMo Data Designer

Octopize sells the Avatar method, which converts confidential tables into synthetic copies, and it is the one vendor in this group that prints its prices. A free first month leads into Starter at €15k, Professional at €48k and Expert at €83k, and the tiers are separated by rows, columns, user seats and annual support days of 2, 9 and 12.

Octopize Mimethik Data (SAS) is registered in Nantes under RCS 839 370 681. The service runs on Scaleway in France, which the page calls HDS certified, or in Docker on your infrastructure. Quotes on the site come from French hospitals, a pharma company and an aerospace group. The vendor says CNIL assessed its automated approach, a claim worth requesting in writing.

Standout feature. Octopize lets a buyer work out the cost of a synthetic data programme from its public pricing page alone.

#2 Syntho

Amsterdam, Netherlands Basic, Standard and Ultimate plans on quote; no consumption charges stated Demo on request

Best for: Developers and QA groups that want production-like copies of several relational databases

• Operating company. Syntho B.V.
• Jurisdiction. EU (Netherlands)
• Where the data sits. Customer's own environment (vendor states it has no access)
• Source code. Closed source
• Replaces. Gretel, NVIDIA NeMo Data Designer

Syntho is an Amsterdam company whose engine mixes three ways of producing data: masking after a scanner finds sensitive values, rules and formulas, and AI models that learn the statistics of production tables. The vendor says the methods can run together in one job, and AI output comes with a quality report.

Syntho B.V. deploys inside the customer's environment and states that it has no access to it. Plans depend on features, database connections and connector types, with no fee per generation, database or workspace and support included. Logos on its site include KPN, Philips, Nederlandse Spoorwegen, Leiden University Medical Center and Statistics Netherlands. On 9 June 2026 it announced the purchase of the MOSTLY AI brand.

Standout feature. Syntho is the tool here that combines masking, formula-built data and AI-learned data in a single run.

#3 Dedomena

Madrid, Spain Start for free; paid pricing not published Demo on request

Best for: Finance and data groups wanting validation, anonymisation and synthetic generation from one vendor

• Operating company. DEDOMENA ARTIFICIAL INTELLIGENCE, S.L.
• Jurisdiction. EU (Spain)
• Where the data sits. Not stated on the pages reviewed
• Source code. Closed source
• Replaces. Gretel, NVIDIA NeMo Data Designer

Dedomena presents synthetic data as one module of a larger platform that also validates datasets, anonymises them, offers a data space for sharing and adds AI agents. Users pick one of four algorithms, one aimed at banking data, and choose which columns to train on.

The vendor is DEDOMENA ARTIFICIAL INTELLIGENCE, S.L. of Madrid, registered at an address in Rivas-Vaciamadrid according to its terms. It states that one generator can learn from several tables or files while preserving referential integrity, and each run closes with a quality report on utility and privacy. Paid pricing and hosting region are not published, and the speed-up figures come from Dedomena itself.

Standout feature. Dedomena is the only platform here where profiling, anonymisation and generation share one workspace with a data-sharing space.

#4 Aindo

Trieste, Italy Founded 2018 See website; no prices published Demo on request

Best for: Hospitals, pharma groups and CROs that need evidence from protected patient records

• Operating company. Aindo SpA
• Jurisdiction. EU (Italy)
• Where the data sits. EU servers (per privacy policy)
• Source code. Closed source
• Replaces. Gretel, NVIDIA NeMo Data Designer

Aindo began in 2018 as an AI lab at the SISSA school in Trieste and now builds its message around public health, real-world evidence, external control arms and clinical intelligence. Synthetic cohorts serve observational studies, and synthetic control arms address trials where recruiting a comparison group is hard.

Aindo SpA lists Europrivacy under Article 42 GDPR, ISO 27001 and ISO 9001, and calls itself the first company in Europe with Europrivacy for synthetic data in healthcare, which is its own description. The privacy policy places processing on EU servers and the platform page says local use is possible. The claim that analysis can begin at once instead of after months of access talks is the vendor's, and no prices appear publicly.

Standout feature. Aindo is the only vendor here that offers synthetic control arms for clinical trials.

#5 MOSTLY AI

Vienna / Amsterdam, Austria Free hosted version at app.mostly.ai; enterprise pricing on request Demo on request

Best for: Data scientists who want to begin with a Python package and add a shared platform later

• Operating company. MOSTLY AI (brand owned by Syntho B.V.; platform operating entity not verified)
• Jurisdiction. EU (Austria)
• Where the data sits. Your own compute with the local SDK; hosted version at app.mostly.ai
• Source code. Closed source
• Replaces. Gretel, NVIDIA NeMo Data Designer

MOSTLY AI is known for its Synthetic Data SDK, which installs with pip and trains a generator on your own CPU or GPU. The home page describes it as a permissive open-source project under Apache v2, and the documentation covers Docker and air-gapped use. A client mode links to a remote platform where locally trained generators can be uploaded.

The SDK models categorical, numerical, geospatial and text columns, in single tables, linked tables and time series, with differential privacy as an advanced training option. The brand was bought by Syntho B.V. on 9 June 2026, and the legal entity running the platform could not be read from public pages, so confirm it before signing. Enterprise prices are on request, while the hosted version at app.mostly.ai is free.

Standout feature. MOSTLY AI lets a data scientist generate a first dataset from a pip install, without any contract.

#6 Syntheticus

Zurich (Dietikon address in terms), Switzerland Founded 2021 See website; no prices published Demo on request

Best for: Swiss and other European companies planning synthetic data for model training, testing or reporting

• Operating company. Syntheticus AG
• Jurisdiction. Switzerland (adequacy decision, outside the EEA)
• Where the data sits. Not stated
• Source code. Closed source
• Replaces. Gretel, NVIDIA NeMo Data Designer

Syntheticus describes itself as a GenAI company in Zurich, founded in 2021, and sorts its use cases into AI and LLM work, software testing, and analytics with business intelligence. For AI projects the vendor claims gains of up to 3 percent in model performance, and up to 5 percent when synthetic data is mixed with real data.

The product terms name Syntheticus AG with an address in Dietikon, while the About page gives Zurich. Switzerland has an EU adequacy decision but is outside the EEA. A blog post announces ISO 27001 and a trademark registration, a trust portal is linked, and a 2023 notice describes a partnership that adds Syntheticus as a data module on the Linebreak edge platform. Prices, a module list and hosting region are absent from the public pages.

Standout feature. Syntheticus is the only vendor here sold under Swiss law with separate offers for training, testing and analytics.

#7 Sarus

Paris, France See website; no prices published Demo on request

Best for: Analysts working on sensitive data they should never see directly

• Operating company. Sarus Technologies SAS
• Jurisdiction. EU (France)
• Where the data sits. Your cloud or on-premise
• Source code. Closed source
• Replaces. Gretel, NVIDIA NeMo Data Designer

Sarus calls itself a privacy layer: data scientists connect to a dataset they cannot read, query rewriting makes each processing job privacy-safe on the fly, and output-level control limits what leaves. Synthetic data is one route among several, made by a generative model trained with differential privacy.

Sarus Technologies SAS is a société par actions simplifiée in Paris, registered as RCS Paris B 879 906 055. It deploys on-premise or in a public cloud with Docker or Kubernetes, and lists PostgreSQL, Snowflake, MongoDB, Tableau, SageMaker and Vertex AI integrations, plus a notice on Azure Confidential Clean Rooms. SarusLLM extends the idea to fine-tuning open-source language models on private data. No prices are shown and the licence of its GitHub repositories was not checked.

Standout feature. Sarus is the tool here that rewrites analysts' queries so they never receive raw records.

#8 K2view

Yokneam, Israel Pricing on request Demo on request

Best for: Banks and telecoms with many linked systems that need consistent test data

• Operating company. K2view Ltd.
• Jurisdiction. Israel (outside the EU/EEA)
• Where the data sits. Not stated
• Source code. Closed source
• Replaces. Gretel, NVIDIA NeMo Data Designer

K2view sells an enterprise data product platform in which synthetic data sits next to test data management, masking and tokenisation. Generation works on whole business entities, so a customer's related records stay consistent across systems, instead of one table at a time.

K2view Ltd. is registered in Israel under number 514314632 at Hayetsira 6, Yokneam, and Israel is outside the EU and EEA. The vendor lists AI generation from masked production patterns, rules-based generation without production data, cloning, masking and subsetting, across applications, files, SQL and NoSQL databases, mainframe and cloud stores. The terms we read cover the website, the privacy policy adds a US address for US users, and the entity signing a product contract needs confirming.

Standout feature. K2view generates complete business entities across connected systems, where the other tabular tools work on tables.

#9 Anyverse

Madrid, Spain Pricing on request; demo and possible trial Demo on request

Best for: Automotive, defence and robotics groups training or validating perception models

• Operating company. Anyverse S.L.
• Jurisdiction. EU (Spain)
• Where the data sits. Cloud engine (region not stated)
• Source code. Closed source
• Replaces. Gretel (visual and sensor data use cases)

Anyverse does not produce tables. Its engine simulates light, materials and sensors, with models for cameras, LiDAR, time-of-flight, radar, RGB-IR, NIR and thermal, and builds scenes procedurally with rain, snow, fog, glare and failure modes included.

Anyverse S.L. is named in its legal notice as registered in the Madrid commercial registry. Rendered layers cover RGB, semantic and instance labels, depth, normals, materials, motion vectors and raw 16-bit images. The site mentions alignment with Euro NCAP and UNECE test protocols and lists TISAX. Pricing is on request and the cloud region is not named.

Standout feature. Anyverse simulates radar, LiDAR and thermal sensors in the same scene pipeline as ordinary cameras.

#10 SKY ENGINE AI

London, United Kingdom Pricing on request Demo on request

Best for: Vision AI engineers in automotive, drones, security and medical imaging who need annotated images

• Operating company. SKY ENGINE Limited
• Jurisdiction. United Kingdom (adequacy decision, outside the EEA)
• Where the data sits. Cloud (region not stated)
• Source code. Closed source
• Replaces. Gretel (image data use cases)

SKY ENGINE AI works with images. Its Synthetic Data Cloud renders scenes and exports training sets with the ground truth attached: semantic and instance segmentation, 2D and 3D boxes, depth maps, normals, object metadata and keypoints for pose work.

The platform simulates visible, NIR, thermal and UWB sensors with physically based rendering and offers GAN-based post-processing for domain adaptation. A domain inspector quantifies the gap between real and synthetic images, and PyTorch and TensorFlow integration plus a built-in code environment support training. The legal notice names SKY ENGINE Limited in London, so the United Kingdom adequacy decision applies. Pricing and deployment beyond the managed cloud are not published.

Standout feature. SKY ENGINE AI includes a domain inspector that puts a number on how far synthetic images differ from real ones.

#11 Clearbox AI

Turin, Italy Founded 2019 See website; no prices published Demo on request

Best for: Italian and European organisations that want a small research team to run synthetic data projects

• Operating company. Clearbox AI Solutions S.r.l.
• Jurisdiction. EU (Italy)
• Where the data sits. Not stated
• Source code. Closed source
• Replaces. Gretel, NVIDIA NeMo Data Designer

Clearbox AI is a Turin team with a research background in machine learning and data privacy, founded on 31 July 2019, with offices in Turin and Milan. The site gives three reasons to use synthetic data: protecting sensitive data, augmenting data for ML and LLM work, and correcting bias.

Clearbox AI Solutions S.r.l. is named in the privacy policy at Corso Castelfidardo 30/a, and the About page cites an in-house CIPP/E certified privacy professional. Much of the newest content concerns Replica Italia, a consumer panel built on card transactions and refreshed monthly, which is a data asset rather than software you run. No self-serve platform, documentation, pricing or hosting details were visible.

Standout feature. Clearbox AI is the project-based option here, run by a team with in-house privacy certification and no product to download.