NinetyFive

The fastest autocomplete

NinetyFive Ultra ($10/month): [Purchase]

• Custom models trained on your codebase
• Priority suggestions
• Local inference, no internet required (coming soon)

NinetyFive Free ($0/month):

• Basic suggestions

Academic licenses? Email us at hello@ninetyfive.gg from your institutional email address.

Choose your favorite editor:

• VSCode Marketplace - Open VSX
• Sublime Text
• IntelliJ
• Neovim
• Emacs

See our docs page for more details on how to get started.

NinetyFive works by training a custom model on your codebase. Combined with an optimized inference engine, NinetyFive is able to provide lightning-fast autocomplete suggestions with 50ms median latency. That's almost as fast as your next keystroke.

All code sent to NinetyFive is subject to a 14-day retention policy and is not used to train any models except your own. Additionally, we do not process binary files, files that have been .gitignore'd, or files that have been processed with git crypt.

Run NinetyFive where your code lives. Our enterprise options focus on secure compute and secure training so you can train models and serve completions inside your own trust boundary.

• Private deployment: with isolated networking and egress controls.
• Data stays private: no cross-tenant training; optional zero-retention or custom retention windows.
• Secure training: run training jobs in isolated nodes with no-egress policies, signed artifacts, and provenance checks.
• Secure inference: stateless completion servers with request logging disabled or redacted; prompts are never used for global training.
• Network options: private endpoints/peering, on-prem connectivity, and data-residency controls.
• Performance: GPU-accelerated inference and training with low-latency paths optimized for IDE completions.

Want a private pilot or security review packet? Email hello@ninetyfive.gg.

Message us on Discord or send us an email at hello@ninetyfive.gg.