Prime Intellect BlogINTELLECT-3 is a 106B parameter expert mixed model, trained on the GLM 4.5 Air base model using SFT and RL, achieving optimal performance at its scale on math, code, science, and reasoning benchmarks, surpassing many larger front-line models.
Dify Blog7070Kakaku has addressed speed, security, and scalability issues in AI application development through the Dify platform, making 30% of employees AI builders and launching a product data extractor within 3 hours, cumulatively building nearly 950 internal applications. Dify Enterprise provides a multi-workspace architecture, single sign-on, sandbox environment, and production-grade infrastructure, achieving a combination of central control and local freedom.
Liquid AI BlogLiquid AI and Shopify have reached a multi-million dollar multi-year agreement, under which Shopify will be authorized to use Liquid AI's flagship Liquid foundation model (LFMs) for search and other use cases
Anthropic ResearchEight Anthropic researchers (with no robotics experience) were randomly divided into two groups, one using Claude and the other not, tasked with using quadruped robot dogs to pick up beach balls. The team using Claude completed more tasks faster and was the only one to achieve the goal of making the robot completely autonomous in picking up the balls. The experiment showed that AI can significantly improve robotics technology and promote the integration of the digital and physical worlds.
Dify BlogDify Knowledge Pipeline has officially integrated TiDB Vector, allowing developers to seamlessly reuse processes like parsing tables, extracting entities, and embedding text for knowledge construction in downstream AI applications. TiDB Vector supports hybrid SQL + vector search, filtering structured metadata via SQL syntax before performing semantic retrieval to ensure precise, context-aware responses, and leverages TiDB's distributed architecture for enterprise-grade scalability and stability.
Dify Blog7070Dify Platform now integrates Qdrant high-performance Rust vector database, providing faster retrieval, hybrid search, and scalable performance for building reliable, production-grade AI applications. This integration enables developers to choose Qdrant without becoming vector search experts through simple configuration, achieving millisecond-level retrieval for millions of documents, complex metadata-based retrieval strategies, and hybrid search capabilities, supporting high-performance scaling from prototype to production.
Prime Intellect Blog7575Prime Intellect is expanding its open-source environment project into a global open evaluation and reinforcement learning environment center, committing millions of dollars in prizes and seeking partners to accelerate the development of open superintelligence. Over the past two months, more than 400 cutting-edge reinforcement learning environments and evaluations have been collected through bounty programs and reinforcement learning residencies. A new large-scale bounty program will expand to thousands of open-source environments, divided into open access and task-only categories, with prize amounts ranging from 100 to 5000 dollars.
Dify Blog6565Bright Data has launched on the Dify Marketplace, bringing real-time structured network data sources to Knowledge Pipeline. Its Web Scraper plugin can be installed and configured in 5 steps, accessing a 72M+ proxy network to collect data from e-commerce, social media, etc., supporting API key authorization, enriching the knowledge base of AI workflows.
Dify Blog6565Dify platform now supports Tavily Real-time Web Search API, allowing users to directly embed web data into knowledge pipelines to enhance accuracy and timeliness; Tavily can be used as a data source plugin in Dify Marketplace, paired with Dify knowledge pipelines, which support modular components from upload and parsing to chunk embedding; users can install the plugin, obtain an API key, and configure LLM nodes to use Tavily search results for generating Q&A pairs and storing them in a knowledge base.
Anthropic ResearchFeatured7171Anthropic has released an open-source auditing tool, Petri (Parallel Exploration Tool for Risky Interactions), where automated agents test target models through multi-round dialogues, and then LLM reviewers score them based on safety dimensions and select the most worthy dialogues for manual review.
Why it matters: Anthropic has open-sourced the automated auditing tool Petri and provided cross-comparison results for 14 cutting-edge models on behaviors such as deception, sycophancy, and self-preservation.
Dify Blog7575The Dify platform allows for rapid, code-free construction of chatbots, AI agents, and intelligent search systems, but deploying in production environments often faces challenges. TrueFoundry AI Gateway seamlessly connects development and production, offering cost control, security reinforcement, service degradation, and supplier switching capabilities, supporting real-time cost tracking, role-based access control, multi-region load balancing, and cross-model compatibility.
Dify Blog7575Dify collaborates with Arize to bring observability tools Phoenix and AX into the Dify ecosystem to evaluate, monitor, and improve agents. Phoenix can automatically track model calls, tool calls, and chain steps, collecting inputs, outputs, latency, and metadata, and supports annotation tracking, building structured test datasets, creating custom evaluations, and running tests.
Prime Intellect BlogEnvironments Hub released, an open community platform for extending RL to Open AGI; supports development, sharing, and reusing RL environments, reducing friction in building, sharing, and reusing environments, promoting open-source AGI progress; now open to everyone, providing RL, Reinforcement Fine-tuning (RFT), computing, and reasoning infrastructure.
Liquid AI BlogLiquid AI announces that its efficient Liquid Edge AI Platform (LEAP) platform now natively supports AMD's latest Ryzen™ and Ryzen AI™ processors, bringing performance improvements for developers building high-quality AI solutions on devices.
Prime Intellect BlogPrime Intellect releases the SYNTHETIC-2 dataset, containing four million verified reasoning trajectories, covering the most comprehensive complex reinforcement learning tasks and validators to date. The dataset is generated through a pipeline parallel distributed reasoning collaboration involving global computing contributors, with 1,253 GPUs (including 49 8xH200 nodes, 43 8xH100 nodes, and a large number of consumer-grade 3090 and 4090 GPUs) participating in its creation over 3 days.
Dify BlogDify v1.6.0 version includes built-in bidirectional Model Context Protocol (MCP) support, allowing direct calls to MCP servers or exposing Dify proxies/workflows as MCP servers to achieve faster and more reliable integration and feature expansion.
Dify BlogDify 1.5.0 is released, enabling workflow debugging through node execution records and variable passing features, supporting single-step execution and real-time variable modification, significantly enhancing the transparency of AI application development.
Prime Intellect BlogDeepSeek releases SYNTHETIC-2, which includes 20 complex reinforcement learning tasks and verification reasoning trajectories, supporting heterogeneous computing from consumer-grade GPUs to super-large-scale clusters, using the DeepSeek-R1-0528 model to generate data and label difficulty.
Liquid AI Blog7575Grafting is a simple two-stage method for exploring new architecture designs by modifying the computational graph of a pre-trained model, including activation distillation and lightweight fine-tuning.
Dify Blog6565Dify has surpassed 100,000 GitHub stars and joined the top 100 global open-source projects. The platform was open-sourced two years ago, aiming to make it easier for developers to build LLM applications, and has continued to grow thanks to community contributions. The project especially thanks 10 core contributors who have driven platform progress through code submissions, feature development, and documentation improvements.
Prime Intellect BlogPCCL released, designed for global distributed training, with fault tolerance, supports operation on the public internet, bandwidth up to 45 Gbit/s between each pair of data centers, training bandwidth up to 25 Gbit/s between North America and Western Europe, supports dynamic member joining and rejoining, verified its high fault tolerance through long-term stress tests on major sockets (Linux, macOS, Windows WSA).
Prime Intellect BlogINTELLECT-2 is the first 32B parameter model trained using global distributed reinforcement learning, employing fully asynchronous RL across dynamic heterogeneous computing contributor groups to train reasoning language models. It has open-sourced the training framework PRIME-RL and components such as TOPLOC and SHARDCAST, while improving the GRPO training recipe and data filtering techniques, enhancing QwQ-32B performance on mathematical and coding tasks.
Prime Intellect BlogPrime Intellect previews its distributed reasoning stack, designed for consumer-grade GPUs and public internet 100ms latency, and releases three open-source code repositories PRIME-IROH, PRIME-VLLM, PRIME-PIPELINE to achieve latency-aware pipeline parallelism, aiming to achieve the openness and democratization of AGI through consumer-grade hardware and network connections.
Dify BlogDify users can create advanced AI chatbots by integrating InfraNodus. The chatbot not only answers questions but also mines insights and generates new ideas from the knowledge base. InfraNodus can visualize the knowledge base as a knowledge graph, identifying 'structural gaps' in concept connections, and its API generates insight-oriented questions targeting these gaps. Users can seamlessly integrate this capability into Dify applications to initiate conversations, content development, or provide smart prompts for Dify AI agents, thereby embedding enhanced chatbots on websites.
Prime Intellect BlogINTELLECT-2 Released, First to Achieve Global Distributed 32B Parameter Model Reinforcement Learning Training, Anyone Can Contribute Heterogeneous Computing Power Without Permission, Eliminates Communication Bottlenecks Through Asynchronous Reinforcement Learning Paradigm, Supports Heterogeneous Inference Nodes, Inference Nodes Only Need 4×RTX 3090 GPUs to Participate in Training.
Dify BlogDify's mcp-server plugin can transform any Dify application into a server-side endpoint compliant with the MCP protocol, allowing external MCP clients to access it directly. The plugin handles client requests via HTTP and SSE protocols, supporting the use of Dify applications as a single MCP tool for Cursor, Claude Desktop, and others, and allows operation in private network environments to enhance security.
Dify BlogDify has launched a free LLM technology stack for the education sector, providing teachers and students worldwide with a complete generative AI toolset from construction to application, supporting drag-and-drop low-code development, visual workflow design, and various plugins. Early users can enjoy a 100% discount coupon.
Dify BlogDify supports integrating over 7000 Zapier applications into AI agents via MCP plugins. Developers can use the MCP SSE plugin to dynamically call external tools or use the Agent Strategy plugin to enable agents to make autonomous decisions on tool calls. The Zapier MCP server provides over 30000 actions, supporting automated tasks like sending emails via Gmail, requiring server endpoint configuration and setting GPT-4o prompts to control calls.
Dify Blog6565Dify's newly released SSH plugin achieves server management through AI-driven automation, capable of automated deployment, file processing, script execution, and troubleshooting. This plugin is developed based on the Cursor tool, supporting SSH remote command execution and file transfer with a context length of up to 8k. It is easy to install and can independently complete server configuration, application deployment, and issue resolution through an AI intelligent agent.
Dify BlogDify integrates Fish Audio plugin from Open Audio into its platform, providing real-time text-to-speech (TTS) and voice cloning functions for AI applications. Fish Audio supports formats such as Opus, MP3, WAV, and cloning voice takes only 30-45 seconds. Users can control output parameters through WebSocket API. Developers can install the plugin on Dify marketplace and configure API keys to automatically convert LLM generated text into audio clips of specified voices.
Dify BlogAgora's conversational AI extension, as a plugin, has launched on the Dify market platform, providing developers with real-time voice AI capabilities. The extension supports multiple LLMs, featuring industry-leading speech-to-text and mainstream text-to-speech services, achieving rapid response and global reliability through Agora's low-latency SD-RTN™ technology. Developers can quickly deploy virtual assistants or customer service robots. The Dify market platform now hosts over 120+ plugins, supporting community sharing and local deployment.
Dify Blog6565Dify.AI will attend the NVIDIA GTC conference from March 17 to 21, 2025, demonstrating the latest innovative achievements in Room 3226, including the plugin system, workflow mechanism, and NVIDIA NIM integration. The core of the company is to help developers define AI applications through prompts, context, and plugins. Version 1.0.0 introduced the plugin mechanism, workflow enhancements, and an open ecosystem.
Dify BlogDify has officially launched the Agent Strategy plugin, enabling workflow nodes to delegate steps outside of fixed processes to LLM for autonomous decision-making. The Agent Node serves as the execution unit, managing resources and reasoning processes; Agent Strategy is a pluggable reasoning algorithm module that defines tool usage and problem-solving logic.
Dify BlogDify v1.0.0 introduces a plugin system, allowing developers to easily integrate real-time search functionality through the Brave Search API plugin. Create smart agents or workflows in Dify Studio, choose models like Claude-3.5 Sonnet, and add the Brave Search tool to enable search. Deploy in workflows or integrate custom plugins to extend functionality.
Dify BlogAfter the release of Dify v1.0.0, its plugin system has become a key feature, achieving module decoupling, a plugin marketplace and sharing mechanism, and Endpoint plugins by separating horizontally scalable modules from Dify and operating them as independent runtimes.
Dify BlogDify achieves multi-step research automation through node iteration, search, and summarization with the DeepResearch application. This application uses LLM to generate queries and summaries, supports an 8k context window, requires DeepSeek and Tavily APIs, and is deployed on the Dify platform.
Dify BlogDify provides three solutions to ensure the stable operation of DeepSeek-R1: official API integration supports multi-key load balancing; access Azure AI Foundry and other MaaS services as backups; locally deploy Ollama to meet privacy requirements. Through Dify, official API and MaaS can be switched between primary and backup to ensure that applications supporting 1000+ models remain online.
.png%3Fv%3D522e7362d902&mode=card&exp=1794873600&sig=f67c6eea92e34cef)
