Does Tabnine train on your data?
Tabnine · AI coding assistant · official site ↗
Doesn’t train on your dataZero-retention availableNo training to opt out of
Does Tabnine train its models on your data?
No — Tabnine does not use your content to train its AI models by default.
Tabnine operates a no-train, no-retain policy: your code is never used to train any of its models and is never stored on its servers — requests are ephemerally processed for suggestions and immediately discarded, regardless of which model is used.
Can you opt out?
There is nothing to opt out of — training on customer code is never done by default. Tabnine's base code and chat models are trained only on permissively licensed open-source code; optional private models can be pretrained on your own code inside your environment and are accessible only to your team.
Zero retention / DPA
Zero code retention is the standing policy: code sent for inference is held only in memory as ephemeral context and deleted immediately after the response is returned. Tabnine offers SaaS, VPC, on-prem and fully air-gapped deployments where no data leaves your infrastructure; backed by SOC 2 Type II, ISO 27001 and GDPR compliance via its Trust Center. DPA ↗
What the listicles get wrong
The nuance content farms miss: the no-train guarantee is genuinely default and model-agnostic, not tier-gated like most rivals. Base models train only on permissively licensed open source; private fine-tuned models live in your environment. Operational metrics/logs (no code, no PII) may be retained ~1 week for support.
Verdict by plan tier
All plans (Dev / Enterprise / self-hosted)No trainingCode is never used to train Tabnine's models and never retained; requests are ephemerally processed and discarded, regardless of model or deployment.
Is Tabnine HIPAA compliant? Does it sign a BAA?
BAA on requestPHI only on covered tier
HIPAA Business Associate Agreement
BAA on request
Tabnine fits HIPAA workflows via its Enterprise self-hosted, VPC, on-prem and air-gapped deployments with zero code retention, where customer code never reaches Tabnine's servers; a BAA is handled through enterprise engagement rather than a self-serve click-through.
Can you put PHI in Tabnine?
Tabnine suits HIPAA use only via its self-hosted/air-gapped Enterprise deployment where you retain full data control — not the default SaaS tier.
https://docs.tabnine.com/main/welcome/readme/security
https://www.tabnine.com/blog/hipaa-cybersecurity-requirements/
Get notified when this changes
We track Tabnine's data-training and retention policy. Leave your email and we'll send one note if it changes.
Frequently asked questions
Does Tabnine train its AI models on my data?
No — Tabnine does not use your content to train its AI models by default. Tabnine operates a no-train, no-retain policy: your code is never used to train any of its models and is never stored on its servers — requests are ephemerally processed for suggestions and immediately discarded, regardless of which model is used.
Can I opt out of Tabnine training on my data?
There is no training opt-out to set for Tabnine: There is nothing to opt out of — training on customer code is never done by default. Tabnine's base code and chat models are trained only on permissively licensed open-source code; optional private models can be pretrained on your own code inside your environment and are accessible only to your team.
Does Tabnine offer zero data retention (ZDR) or a DPA?
Zero code retention is the standing policy: code sent for inference is held only in memory as ephemeral context and deleted immediately after the response is returned. Tabnine offers SaaS, VPC, on-prem and fully air-gapped deployments where no data leaves your infrastructure; backed by SOC 2 Type II, ISO 27001 and GDPR compliance via its Trust Center.
Is Tabnine safe to use with confidential or proprietary data?
It depends on your plan tier. All plans (Dev / Enterprise / self-hosted): Code is never used to train Tabnine's models and never retained; requests are ephemerally processed and discarded, regardless of model or deployment. Always confirm current terms with Tabnine before sending confidential data — this is cited public information, not legal advice.
Is Tabnine HIPAA compliant? Does it sign a BAA?
BAA on request. Tabnine fits HIPAA workflows via its Enterprise self-hosted, VPC, on-prem and air-gapped deployments with zero code retention, where customer code never reaches Tabnine's servers; a BAA is handled through enterprise engagement rather than a self-serve click-through. Confirm a current BAA directly with Tabnine before sending PHI — this is cited public information, not legal or compliance advice.
Can I put PHI (protected health information) in Tabnine?
PHI only on covered tier. Tabnine suits HIPAA use only via its self-hosted/air-gapped Enterprise deployment where you retain full data control — not the default SaaS tier.
Sources
https://www.tabnine.com/code-privacy/
https://docs.tabnine.com/main/welcome/readme/privacy
This page is cited public information, not legal or compliance advice. Whether Tabnine trains on your data, and any zero-retention or DPA option, can depend on your specific plan, region and contract. Always confirm current terms with Tabnine before sending confidential or proprietary data.