Which AI models does Lovable use?

A short technical overview of the AI providers and models that power Lovable, and how to check whether they're currently healthy.

The short answer

Lovable utilizes frontier models from three providers: OpenAI, Anthropic, and Google Cloud. The flagship models commonly used are GPT-4o (OpenAI), Claude 3.5 Sonnet (Anthropic), and Gemini (Google Cloud). The right model is selected based on the task — code generation, planning, multimodal input, or fast edits.

Per provider

OpenAI — GPT-4o

Strong general-purpose reasoning and code generation, with multimodal (image) input. Used for a large share of build and edit operations.

Anthropic — Claude 3.5 Sonnet

Excellent at long-context reasoning and careful code edits across larger codebases. Often picked for higher-stakes refactors and planning.

Google Cloud — Gemini

Used for multimodal tasks and selected workloads that benefit from Gemini's long context window and Google Cloud's infrastructure.

Live status of these providers

Outages at any of these providers can affect Lovable's responsiveness. The Lovable Status dashboard aggregates the official status feeds for OpenAI, Anthropic, and Google Cloud (plus the other platforms Lovable depends on) and refreshes every 2 minutes.

FAQ

Can I choose which model Lovable uses?

Lovable routes requests automatically based on the task. Model selection is managed by the platform.

Does Lovable train on my code?

No — Lovable uses provider APIs under their standard no-training terms. See Lovable's own documentation for the authoritative policy.