Ecosystem & Applications

Real-world deployments and applications of LongCat AI

Product Applications

LongCat App

Multi-modal AI assistant supporting voice, image, and file interactions. Provides real-time conversation and task assistance. AI image generation powered by LongCat-Image: generate high-quality images from text prompts, or edit generated images with multi-round iterations — all with fast response and studio-quality output.

  • Text-to-image: Generate high-quality images from text prompts with superior Chinese text rendering (covers all 8,105 standard Chinese characters)
  • Image-to-image: Upload any material (landscape photos, selfies, sketches) and generate new images based on requirements
  • 24 zero-threshold templates: Covering poster design, portrait refinement, scene transformation, and more — click "AI Creation" to use directly, no prompt anxiety
  • Image editing: 15 editing task types with natural language (open-source SOTA on GEdit-Bench, ImgEdit-Bench)
  • Multi-modal interaction: Voice, image, and file uploads
  • Available on: LongCat Web and LongCat APP (iOS: search "LongCat" in App Store)

NoCode AI Coding Assistant

Rapid development tool powered by LongCat models, enabling AI-assisted coding and development workflows.

"Xiaomei" Agent

Everyday AI life assistant integrated into Meituan's services, providing intelligent support for daily tasks.

Internal Deployments

Meituan has deployed LongCat models across multiple internal services:

  • AI-assisted coding tools: Development productivity
  • Intelligent meeting transcription: Meeting summarization
  • Automated documentation generation: Content creation
  • Customer service automation: Support efficiency

Models are optimized for domain-specific use cases including meal delivery, travel booking, and more.

Embodied Intelligence Investments

LongCat AI serves as the cognitive core for Meituan's embodied intelligence strategy:

  • Stone Robotics: Service robots
  • Xinghai Map: Navigation and mapping
  • Autonomous delivery: Ground and low-altitude networks

Deployment & Integration

LongCat models support various deployment scenarios with specialized optimizations:

  • SGLang and vLLM: High-throughput inference
  • Cost-efficient inference: ~$0.7 per 1M output tokens on H800 GPUs
  • Scalable architecture: Single-user and multi-user scenarios
Deployment Guide Explore Models