Search results for "GPU"

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.

2026-07-02

Dabing AI Voice Changer is a professional real-time AI voice-changing tool featuring ultra-low latency (under 100ms) and a library of over 500 meticulously tuned voices. It requires no dedicated GPU and consumes only about 5% of CPU resources, allowing it to run smoothly even on standard laptops. Perfectly suited for scenarios such as multiplayer gaming, live streaming interactions, and social voice chats, it makes transforming your voice simple and fun.

2026-07-11

Supertone Shift is a real-time AI voice-changing software developed by Supertone (a subsidiary of HYBE). It is renowned for its industry-leading low latency—as low as 47ms—and a lightweight design that operates without the need for a dedicated GPU. The software offers a vast library of AI voice personas created by professional voice actors and supports fine-tuned adjustments to parameters such as pitch, dynamic range, and reverb. Ideal for scenarios ranging from multiplayer gaming and VTuber streaming to video voiceovers, it is highly popular among creators worldwide, particularly in Japan.

2026-07-11

Memo AI is an all-in-one, local audio-to-text tool that effortlessly converts content from YouTube, podcasts, and local files into transcripts. It offers features such as transcription and translation for over 90 languages, AI-generated summaries, AI mind maps, and text-to-speech capabilities. Operating entirely locally to ensure privacy, it supports Windows and macOS and leverages GPU acceleration for M-series chips, enabling a 30-minute audio file to be transcribed in just two minutes.

2026-07-11

Tabby is an open-source, self-hosted AI coding assistant designed to bring the power of AI to your development workflow while keeping you in control. Whether coding in the cloud or on-premises, Tabby offers a flexible, transparent, and highly configurable alternative to proprietary solutions. It features cutting-edge AI capabilities for code completion, answering queries, inline chat, and more, easily integrates with existing infrastructure including Cloud IDEs, supports consumer-grade GPUs, and requires no external DBMS or cloud services, ensuring software supply chain safety.

2026-07-17

PaddlePaddle AI Studio is an AI learning and training community under Baidu, designed for AI learners and developers. It integrates free AI courses, deep learning sample projects, classic datasets, and provides cloud-based powerful GPU computing and storage resources. Users can participate in beginner challenges and elite algorithm competitions, explore large model communities and applications, enabling end-to-end learning and development from entry to practice.

2026-07-22

Kaggle is the world's largest data science community and machine learning competition platform. It offers a vast repository of public datasets, GPU-accelerated Jupyter Notebooks, and prediction modeling competitions hosted by companies and research institutions. Users can learn data science online, tackle real-world data challenges, collaborate with global data scientists, improve model accuracy, and win prizes. Ideal for both beginners and experts, Kaggle is the go-to platform for practicing and showcasing data analysis and deep learning skills.

2026-07-22

Unsloth is an open-source tool designed to accelerate fine-tuning of large language models, supporting Llama, Mistral, Gemma, and more. With double quantization and optimized kernels, it achieves up to 2x speed improvement and 80% memory reduction on consumer GPUs while maintaining model accuracy. No need for high-end hardware; free to use on Colab, ideal for individual developers and SMEs to quickly customize AI assistants.

2026-07-26

Meituan has officially open-sourced LongCat-2.0, a massive MoE model with 1.6 trillion parameters. This release achieves end-to-end training and inference deployment on a domestic computing cluster comprising 50,000 GPUs, thereby breaking reliance on overseas computing power. The announcement details LongCat-2.0's technical highlights, performance advantages, and industry value.

Introduction: Why Does Your AI Art Always Miss the Mark? Hey folks, we all know the most frustrating part of AI art isn't struggling with the model or your GPU overheating—it's this: using the exact ...