Lovable Now Lets Builders Swap GPT-5.6 and Gemini 3.6 Flash Inside Apps

Lovable, the AI-powered app builder that lets you ship full-stack web apps through conversation, just expanded the model menu inside its AI Connector. You can now power your app's AI features with Gemini 3.6 Flash or any of the three GPT-5.6 variants: Sol, Terra, and Luna. Previously, the Connector was locked to a single Gemini model with no switching option.

What the AI Connector actually does

Lovable has two distinct AI layers. The Lovable agent is what you talk to when building your app. The AI Connector is a runtime service your deployed app's users interact with directly. When a customer uses the chatbot or document summarizer you built, that's the Connector at work. The model you pick here determines the quality and cost of every AI interaction your users have, whether that's a chatbot, a document summarizer, or a sentiment analysis tool. You enable it from Settings, then Connectors.

The new models, explained

Both model families landed recently. Gemini 3.6 Flash launched with 17% fewer output tokens and lower pricing than its predecessor, 3.5 Flash. That efficiency compounds in agent loops: you pay a lower rate on a smaller token count to complete the same tasks.

GPT-5.6 comes in three tiers ranked from least to most capable:

• GPT-5.6 Luna — fast and low-cost, designed for high-volume workflows.
• GPT-5.6 Terra — competitive with GPT-5.5 at half the price.
• GPT-5.6 Sol — OpenAI's self-described "workhorse" and "best coding model yet," built for complex reasoning, coding, and agentic workflows.

How Gemini 3.6 Flash and GPT-5.6 compare

These two families take different approaches. Gemini 3.6 Flash generates 275.5 tokens per second versus GPT-5.6 Sol's 60.4, a significant speed advantage. On coding benchmarks, Gemini 3.6 Flash trails GPT-5.6 Luna and Grok 4.5, but leads on long-context retrieval and efficiency gains over its predecessor.

Gemini also handles a wider range of inputs: text, image, video, audio, and PDF. GPT-5.6 across all tiers accepts only text and images, making Gemini the only option if your app needs to process video, audio, or PDF files.

One underrated upgrade in Gemini 3.6 Flash: the knowledge cutoff advances from January 2025 to March 2026, a 14-month jump that matters more for everyday users than any single benchmark score.

Picking the right model for your app

• High-volume chatbots or real-time features — Gemini 3.6 Flash or GPT-5.6 Luna. Both are fast and cost-efficient.
• Document processing, PDFs, or video understanding — Gemini 3.6 Flash. GPT-5.6 requires separate tooling for those modalities.
• Complex coding agents or reasoning-heavy tasks — GPT-5.6 Sol, which posts state-of-the-art results across coding, knowledge work, cybersecurity, and science.
• Balanced everyday work at lower cost — GPT-5.6 Terra, which matches GPT-5.5 quality at half the price.

How to access it

The new models are available now. Navigate to Connectors, then AI inside your Lovable project to see the full model list. No API keys or external accounts are required for the managed Gemini option. Lovable bills AI Connector usage against your workspace's AI balance, separate from the credits used to build the app itself.

For teams building AI-native products on Lovable, having Gemini 3.6 Flash's long-context and multimodal strengths alongside GPT-5.6's coding and reasoning ceiling in the same connector menu means you can match the model to the job rather than working around a fixed default.