Lovable has added new AI models and tools that let builders create apps for tutoring, research, support triage, and video creation, all without leaving the platform. The update brings GPT-Live from OpenAI, the latest Claude models from Anthropic, Gemini 4 Argon, Gemini Omni, and Lyria from Google, and data tools from Perplexity, Firecrawl, Parallel, and Jev by Typesafe AI.

Real-time voice from GPT-Live

GPT-Live gives users a more natural way to interact with an app. Developers can build voice assistants that answer questions, navigate features, and complete tasks, from booking an appointment to getting customer support, all through real-time conversation. The article suggests three examples: a customer assistant that books appointments, a language tutor that helps students practice speaking before an exam, and a sales pitch coach that asks follow-up questions and helps users refine their pitch. A Lovable team member built a demo in which users can talk to Rocky from Project Hail Mary.

Reasoning and generation with Claude, Gemini, and Lyria

Claude is positioned for work that requires deep reasoning, careful planning, and polished output. Lovable says apps can now carry a task from the first step to the finished result. Examples include a tool that turns client briefs into finished, well-designed proposals, an app that takes a photo of a marketplace listing and estimates what the item is really worth, and an inbox assistant that flags urgent messages and drafts replies.

On the media side, Gemini Omni, Google's latest video model, can animate photos or create videos from any input. Lyria writes original tracks with lyrics in the style, tempo, and vocals the builder chooses. Gemini 4 Argon covers supporting knowledge work. Possible apps include a shop tool that turns one product photo into a promo video with its own music, a real estate app that turns listing photos into a walkthrough clip, a yoga studio app that gives every class its own soundtrack, and a market research app that reads competitor reports and trend charts and writes the brief for the next campaign.

Live web answers and clean data

Perplexity gives apps real-time answers grounded in the live web, with citations users can check. The article suggests a morning briefing for investment teams that reads overnight news, and a city guide that answers “what's on this weekend?” with links to every event.

Firecrawl, described as “the knowledge library for AI,” lets apps search, scrape, and pull clean data from anywhere. Suggested builds include a price tracker that watches competitor pages and a help-center assistant that answers from a company's own docs.

Parallel brings the live web to every app through an AI web search platform. It can handle quick facts, such as what happened in the news yesterday, and deep research, such as mapping a company's investments over decades. Suggested tools include a competitive news and intelligence tool, a lead enrichment tool, a research paper monitor, a fundraise tracker, and a grants and RFP finder.

Fast decisions with Jev

Jev by Typesafe AI produces decisions an app can act on, such as a category, a score, or a yes or no. A live chat could tag every message as a refund, a delivery question, or urgent, and a support queue could route each ticket to the right team. Paired with Claude, Jev sorts incoming requests while Claude writes the reply.

Key Features at a Glance

• GPT-Live voice assistants: Real-time conversation for answers, bookings, and tasks – users can talk to the app instead of navigating menus.
• Latest Claude models: Reasoning and multi-step output for proposals, valuations, and inbox drafting.
• Gemini Omni video: Animates photos or generates video from any input – turns a single product photo into a promo clip.
• Lyria music: Original tracks with lyrics in the chosen style, tempo, and vocals – gives each piece of content its own soundtrack.
• Perplexity answers: Live-web responses with citations – users can verify what the app tells them.
• Firecrawl data: Search, scrape, and clean data from any source – powers price trackers and docs-based assistants.
• Parallel research: Quick facts through multi-year deep research – supports competitive intelligence and fundraise tracking.
• Jev classification: Categories, scores, and yes/no decisions the app can act on – routes tickets without manual sorting.

Availability

The article gives no release date or pricing. Builders can describe the feature they want in plain language, and Lovable selects a model or tool. They can also name one in the chat, such as “use Opus 5.5” or “use Perplexity.”