Training

Training

SenseChat is an AI life assistant developed by SenseTime, specializing in chat-based Q&A and content creation. It excels at summarizing financial news, interpreting policies, analyzing earnings reports, and assisting with copywriting, image generation, coding, and engaging conversations on various topics. From writing Xiaohongshu notes and generating PPT outlines to English conversation practice, programming help, poetry creation, children's book illustration, Emoji expression, and logo design, SenseChat offers diverse intelligent services to make life and work easier.

Training

Tianrang Intelligent is an enterprise-level AI decision intelligence platform, offering end-to-end solutions from data collection and model training to intelligent decision-making. Core features include AI applications for city brain, financial risk control, and smart retail, supporting real-time data analysis, predictive modeling, and automated decision-making. Suitable for government, finance, retail and other industries, helping enterprises achieve intelligent transformation.

Training

Miracle Vision is an AI-powered video generation and editing tool designed for content creators, marketers, and video enthusiasts. It leverages advanced AI technology to quickly generate high-quality videos from text, images, or video footage, and offers intelligent editing, special effects, voice synthesis, and subtitle generation. Users can produce professional-grade video content in minutes without professional video editing skills. It is suitable for social media short videos, promotional videos, educational training videos, and more, greatly improving creative efficiency.

Training

Evidently AI is an open-source ML model monitoring and data quality evaluation platform focused on detecting model performance degradation, data drift, and concept drift in production. It supports various scenarios from tabular data to NLP and recommendation systems, offering visual dashboards, automated report generation, and real-time alerts. Helps data scientists and ML engineers quickly identify root causes of model decay, ensuring continuous reliability of AI systems.

Training

AGPT is an AI-powered code generation and collaborative development platform designed to help development teams quickly build, iterate, and deliver software projects. It integrates intelligent code completion, automated code generation, project template management, and team collaboration features, supporting multiple programming languages and frameworks. By describing requirements in natural language, AGPT can generate high-quality code snippets, functions, and even complete modules, significantly improving development efficiency. Suitable for agile development, prototyping, rapid iteration, and especially beneficial for small to medium-sized teams and individual developers.

Training

Mobvoi Open Platform is a one-stop AI capability platform launched by Mobvoi, offering a range of artificial intelligence services including speech recognition, semantic understanding, text-to-speech (TTS), and wake-up word customization. Designed for developers, enterprises, and partners, the platform supports rapid integration of intelligent voice interaction features, widely used in smart hardware, in-car systems, smart home, and customer service robots. With rich APIs and SDKs, it helps users lower the barrier to AI development and achieve efficient, stable voice interaction experiences.

Training

AgentGPT is a powerful autonomous AI agent creation tool that allows users to configure and deploy custom AI agents through a simple web interface. No programming experience is required; users only need to name the target AI agent and set goals. AgentGPT will automatically break down tasks, execute operations, and iterate optimizations until the goal is achieved. The tool supports a variety of use cases, such as workflow automation, content generation, data analysis, customer service, etc., making it suitable for individual developers, enterprises, and teams to quickly implement AI-driven automation solutions. Its core advantages include zero-code operation, autonomous task planning and execution, and scalable agent behavior.

Training

Coze is an all-in-one AI bot development platform launched by ByteDance, enabling users to quickly build, debug, and deploy intelligent agents without coding. It features a rich plugin marketplace, knowledge base management, workflow orchestration, and long-term memory capabilities. Bots can be published across multiple platforms (e.g., Feishu, WeChat, Web). Ideal for customer service bots, content generation assistants, educational tutoring, and office automation, significantly lowering the barrier to AI application development.

Training

Gradio is an open-source Python library designed for machine learning engineers and data scientists to quickly create interactive model demo web interfaces. Without front-end development experience, you can encapsulate a model into a web application with input controls (such as text boxes, image uploads, sliders, etc.) and output displays in just a few lines of code. It supports local operation and one-click deployment on platforms like Hugging Face Spaces, greatly accelerating model sharing, testing, and collaboration.

Training

Jan is a free and open-source AI assistant tool designed to run large language models locally on user devices without requiring internet connectivity. It supports multiple mainstream models (e.g., Llama, Mistral) and offers a clean, intuitive interface for easy model loading and management. Jan prioritizes privacy by processing all data entirely on-device, making it ideal for users concerned about data security. Suitable for personal learning, writing assistance, and lightweight development tasks, Jan delivers efficient and private AI experiences.

Training

Cohere is an enterprise-focused AI large language model platform that provides powerful natural language processing (NLP) capabilities, including text generation, semantic search, classification, and retrieval-augmented generation (RAG). Its key advantages include support for private deployment, data security and control, and optimized model accuracy and efficiency for enterprise use cases. It is widely used in intelligent customer service, knowledge base retrieval, document analysis, content generation, and more, helping enterprises quickly build LLM-based intelligent applications.

Training

Chatbox AI is a cross-platform AI desktop client tool supporting Windows, macOS, and Linux. It integrates multiple mainstream AI models (such as GPT, Claude, Gemini, etc.), offering features like smart chat, code generation, text translation, and document analysis. Users can interact with AI through a clean interface, making it suitable for programming assistance, study research, daily office work, and more. The tool supports offline use to protect user privacy and provides custom API access for advanced users.

Training

iFlytek AI University is an AI professional learning and training platform under iFlytek's open platform. Leveraging iFlytek's technological expertise in NLP, intelligent speech, and other fields, it offers a full-chain course from beginner to advanced model fine-tuning and agent development. The platform is deeply integrated with iFlytek's MaaS tools, supporting zero-code model customization and deployment. Upon completion, students receive official certification and it serves as a core hub for offline developer communities in multiple cities across China.

Training

GPT-4o (“o” stands for Omni) is OpenAI's next-generation flagship multimodal large model, released in May 2024. It enables real-time inference across text, audio, images, and video, with an audio response time as fast as 232 milliseconds, approaching human conversational reaction speed. It possesses groundbreaking capabilities such as emotion perception, real-time translation, and image generation. Its API is twice as fast as GPT-4 Turbo, costs only half the price, and is available to all free ChatGPT users.

Training

ModelScope is a leading open-source AI model platform in China, initiated by Alibaba DAMO Academy in conjunction with the CCF Open Source Development Committee. Adhering to the "Model as a Service" (MaaS) concept, the community has gathered over 200,000 open-source models, 30 million developers, and over 500 contributing organizations. It provides end-to-end services from model experience, download, fine-tuning, training to deployment, covering LLM, multimodal, speech, and AIGC fields. As a core infrastructure for AI developers in China, it is promoting the democratization of AI technology through "open source."

Training

PaLM 2 is Google's next-generation large language model, introduced at Google I/O 2023. It boasts enhanced multilingual capabilities (covering over 100 languages), logical reasoning and mathematical abilities, and code generation capabilities, offering four versions ranging from mobile-friendly Gecko to high-performance Unicorn. PaLM 2 has been integrated into over 25 Google products, including Bard, Google Workspace, Med-PaLM 2, and Sec-PaLM, serving as a core model driving Google's AI strategy.

Training

Replicate is a cloud platform that allows developers to run, fine-tune, and deploy open-source AI models through a simple API. It offers thousands of community-contributed models covering areas such as image generation, large language models, and audio/video processing. No infrastructure management is required, and billing is based on actual computing power used. New users receive a free $25 credit.

Training

Gemma is an open-source AI model family built by Google DeepMind based on Gemini technology. It offers various parameter scales from 2B to 31B, supports multimodal understanding including text, images, and audio, and supports over 140 languages. It is licensed under the Apache 2.0 license and can be used commercially for free. Known for its lightweight and high performance, Gemma can run locally on consumer hardware and has been downloaded over 400 million times, making it a mainstream choice for developers building AI applications.

Training

The Meta AI Developer Platform is the core entry point for Meta's generative AI ecosystem. Through the open-source Llama model, the high-performance Muse Spark programming model, and the Meta Model API, it provides developers with end-to-end support, from model fine-tuning and application building to commercial deployment. Its core strategy is to evolve from "content-generating AI" to "task-execution AI Agents," aiming to make AI the infrastructure driving product efficiency and commercial transformation.

Training

Hugging Face is a leading global AI open-source community and model hosting platform, widely hailed as the "GitHub of machine learning." It provides core services such as model libraries, datasets, and Spaces application hosting, aggregating over 2.7 million models, 950,000 datasets, and 1.2 million AI applications. Through open-source tool libraries like Transformers, the platform significantly lowers the barrier to AI development, becoming an indispensable infrastructure for modern AI developers.

Training

M6 is a general-purpose multimodal large model developed by Alibaba DAMO Academy. In October 2021, it became the world's first AI pre-trained model with 10 trillion parameters. Based on the self-developed Whale distributed framework, it was trained in just 10 days using only 512 V100 GPUs, with energy consumption only 1% of GPT-3. It supports tasks such as image and text generation, visual question answering, and poetry creation, and has been deployed in more than 40 scenarios, including Tmall virtual anchors and Taobao search. It is the core predecessor of the Tongyi Thousand Questions large model.

Training

StableLM is an open-source large language model series launched by Stability AI. It is trained on The Pile extended dataset, which contains 1.5 trillion tokens. It offers multiple parameter versions, including 1.6B, 3B, 7B, and 12B, and supports text generation and code writing. It is licensed under the CC BY-SA 4.0 open-source license, allowing free commercial use. Stable LM 2 12B outperforms Llama 2 70B in some benchmark tests. It is suitable for developers, researchers, and small and medium-sized enterprises for private deployment.

Training

LLaMA is an open-source family of large language models launched by Meta, evolving from LLaMA 1 in February 2023 to multimodal LLaMA 4. The 70B parameter version boasts performance close to GPT-4, while the 405B version rivals GPT-4o. It is completely open-source and commercially viable. Enterprises can download and deploy it on their own servers without data sharing, meeting compliance requirements in industries such as finance and healthcare. With over 1.2 billion downloads globally, it is suitable for developers and enterprises requiring private deployment and deep customization.

Training

Tencent Hybrid is an industrial level AI model independently developed by Tencent. After the reconstruction of pre training and reinforcement learning infrastructure in February 2026, it has been repositioned with "pragmatism" as its core. The Hy3 preview, released in April 2026, is a 295 billion

Training

OpenAI Codex is an AI programming agent launched by OpenAI, with a cloud based software engineering agent as its core form, capable of independently completing the entire development process from requirement understanding to code delivery. After users describe the task objectives through natural

Training

ERNIE is an industry level knowledge enhancement model independently developed by Baidu. Starting from the release of version 1.0 in 2019, it has undergone multiple generations of technological iterations and has established a three-level system of basic model, task model, and industry model. The