SuperAPI AI Directory
Partner with us中文AI newsFavorites

All AI tools

35,702 results · Page 951 of 992

LongCat-Next

AI developer platforms

LongCat-Next is an open-source discrete native multimodal foundation model from Meituan's LongCat team, unifying text, vision, and audio processing under a single autoregressive objective. Built on the LongCat-Flash-Lite MoE backbone (A3B scale), it proposes the DiNA unified paradigm, extending next-token prediction from language to native multimodality and internalizing different modalities into a shared discrete token space. The model comes with an open-source tokenizer; on the vision side, it employs a Semantic Alignment Encoder (SAE) and Residual Vector Quantization (RVQ) to construct hierarchical discrete tokens, and proposes dNaViT as a flexible and unified visual discrete interface. Its visual understanding and generation are reframed as two manifestations of the same prediction process, maintaining strong generation quality even at a 28x compression ratio, with particular strength in text rendering, while also excelling in speech understanding, low-latency speech dialogue, and customizable voice cloning. The code and weights are released under the MIT license, and supervised fine-tuning (SFT) code based on PyTorch FSDP2 is provided. Usage requires at least 3 GPUs with 80GB of VRAM.

Read more

LongLive

AI video

LongLive is a long video generation and world action model infrastructure project released by NVIDIA NVlabs, covering research related to video generation, world models, and embodied world action models. Each subproject in the repository is relatively independent, containing its own code, documentation, and model weights, and the desired projects can be downloaded on demand via sparse checkout. The main directions include: real-time interactive long video generation (LongLive 1.0), NVFP4 parallel training and inference infrastructure for long video generation (LongLive 2.0), a general distillation method (LongLive-Plug), a long-context world action model (Long-WAM), a retrieval-augmented long video generation framework (LongLive-RAG), and more. The project provides training and inference code, public model weights, benchmark evaluations, and real-machine deployment examples, and is released under the Apache License 2.0.

Read more

Loomly

AI marketing

Loomly is a social media management platform for marketing teams, offering features such as content ideation, post creation, scheduling and publishing, engagement management, and data analytics, with an AI system running through the entire workflow. It can generate content suggestions based on the brand, audience, and past performance, build a social media calendar, generate post copy, images, and variations, assist team collaboration and approvals, help safely handle comments and direct messages, and track mentions, trends, and competitors. It officially offers a 7-day free trial with no credit card required.

Read more

LoomVid

AI video

LoomVid is a narrative-driven AI visual storytelling platform that weaves text, photos, and videos into cinematic visual stories. It offers free tools such as text-to-image and image-to-video, and comes with over 14 editing tools and over 437 visual styles built in, helping users turn photos into cartoon characters, generate character portraits, or transform text prompts into dynamic video narratives. The platform is aimed at content creators, designers, brand strategists, educational content producers, film and television creators, and others, enabling them to perform background removal, object erasure, old photo restoration, outfit changes, hairstyle changes, facial expression changes, and more without any design experience. The company states that it supports 4K output and offers a free tier and paid plans; specific pricing and feature details are subject to the official pricing page.

Read more

LoongFlow

AI developer platforms

LoongFlow is an expert-level Agent development framework open-sourced by Baidu, oriented toward "loop engineering." Through the Plan-Execute-Summary (PES) paradigm and structured experiential memory, it enables AI to continuously think, execute, reflect, and evolve in complex software engineering, mathematics, and machine learning tasks. With the philosophy of "the unity of knowledge and action," its goal is to enable developers to quickly transform professional knowledge into expert-level Agents.

Read more

Lotiqlab

AI video

Lotiqlab is a free online Lottie animation editing and motion tool that runs entirely locally in the browser, with no uploading or conversion servers required. It supports creating, editing, batch recoloring, watermark removal, and exporting to GIF, MP4, PNG, dotLottie, and other formats. Its AI mode can generate editable Lottie animations directly from text prompts, without a timeline or After Effects. The tool is aimed at product teams, offering personal and team workspaces, role-based permission management, free use of the editor and converter, and a Pro version ($10/month) that unlocks high resolution, long-duration exports, WebM, the optimizer, and other features.

Read more