Dimension scores are derived from public data and fields; weighted into the composite. Reference only.
NinetyFive is an AI code autocompletion service operated by Numerata Inc, positioned as βthe fastest autocomplete.β Its public materials say it uses an optimized inference engine to deliver completions with 50ms median latency. Its core use case is in-IDE code suggestions, while Ultra users can also train custom models based on their own codebases.
In terms of functionality, NinetyFive focuses on code completion rather than a full IDE or Agent workflow. It supports editors including VSCode Marketplace, Open VSX, Sublime Text, IntelliJ, Neovim, and Emacs, giving it fairly broad coverage. The VSCode example shows that ninetyfive.backendURI can be configured, with login handled via an API key. On the data side, code is subject to a default 14-day retention policy and is not used to train any models other than the userβs own custom model. It also does not process binary files, files covered by .gitignore, or files handled by git crypt.
Pricing is straightforward: Free costs $0/month and provides basic suggestions; Ultra costs $10/month and includes codebase custom models and priority suggestions, while local inference is still marked as coming soon. Academic licensing requires email contact. The enterprise offering is one of its highlights, with support for private deployment, isolated networks, outbound controls, optional zero retention or custom retention periods, isolated training nodes, signed artifacts, inference services with no logs or redacted logs, private endpoints/peering, on-prem connectivity, and data residency controls.
Its strengths include broad editor ecosystem support, low pricing for individual users, clear disclosure of its code retention policy, and relatively detailed enterprise security descriptions. The downsides are that the public materials do not specify which programming languages or frameworks are supported, and the API documentation lacks detail. Local inference is not yet officially available. The terms of service also state that availability is not guaranteed, and AI-generated completions still need to be reviewed and tested by the user.
NinetyFive is suitable for individual developers who want lightweight, low-latency AI completion, as well as teams that need private codebase-based training. Enterprises with strict security requirements may want to look into its private deployment option, but should request a security review package for further evaluation. The public materials do not provide information on access from mainland China, and payment methods are not disclosed. If network or compliance constraints are a concern, it may be worth comparing with GitHub Copilot, Codeium, Tabnine, Amazon Q Developer, JetBrains AI Assistant, or Cursor.
β This review is compiled from public sources and does not constitute a purchase recommendation. Verify all facts on the vendor's official site. Verify on ninetyfive.gg official site.
ninetyfive.gg is an overseas Dev Tools provider. TG4G tracks its product information, an overall rating of 7.0/10, and a China-accessibility score of Workable. Click "Visit Official Site" to reach ninetyfive.gg directly.