Vectoree

Vectoree's pitch is that a coding agent can write the interface and business logic of an app, but still needs somewhere to call models, log users in, store data and keep files. Vectoree puts those pieces behind one project key. The model gateway is OpenAI-compatible, so an app keeps using the OpenAI SDK with a different base URL, and it reaches models from xAI, OpenAI, Google, Anthropic and DeepSeek, along with Vectoree's own vectoree/auto, which the homepage describes as half the price of Flash. An MCP tool hub exposes search and SEO tools such as Similarweb, Semrush and Ahrefs through the same key.

The other blocks are a sign-up and sign-in service for your app's users, one Postgres database per project with filtering by who is signed in, and storage buckets you can make public or private. Templates offer project skeletons whose prompts you copy into your agent.

The workflow is built for agents. You give the coding agent the Vectoree skill file, then log in and link a project with the @vectoree/cli, and the agent runs CLI commands for the database, storage and AI calls. A one-command script switches Codex CLI or the ChatGPT desktop app to use the Vectoree gateway as its model provider, with instructions for macOS, Linux and Windows.

There's no subscription. Creating a project is free, linking the CLI adds $5 of welcome credit, and after that you top up an organisation wallet through Stripe. Gateway calls deduct each model's catalog price, an empty wallet blocks further calls, and usage appears under the organisation's billing page.

Vectoree is run by VECTOREE INC. in Singapore. Its privacy policy, updated 23 October 2026, says it acts as a controller for account and billing data and as a processor for the data your projects store, under a data processing addendum where one applies. Prompts and files are processed to fulfil requests according to your project's retention settings, and data is shared with model and integration providers only to run the features you call.