Blog chia sẻ kiến thức FPT Cloud

In the AI era, leveraging large language models (LLMs) to enhance programming productivity is becoming increasingly common. Instead of relying on expensive international services, developers in Vietnam now have access to FPT AI Marketplace — a domestic AI inference platform offering competitive pricing, high stability, and superior data locality.  This article provides a step-by-step guide to integrating FPT AI Marketplace’s model API into Cursor IDE, enabling you to utilize powerful code generation models directly within your development environment.  1. Creating an FPT AI Marketplace Account Visit https://marketplace.fptcloud.com/ and register for an account.  Special Offer: New users will receive $1 in free credits to experience AI Inference services on the platform! 2. Browse the List of Available Models After logging in, you can view the available models on FPT AI Marketplace.  Figure 1: List of available models on FPT AI Marketplace  For optimal code generation results, it is recommended to select models such as Qwen-32B Coder, LLaMA-8B, or DeepSeek.  3. Generate an API Key Please log in and navigate to https://marketplace.fptcloud.com/en/my-account#my-api-key        Click “Create new API Key”, select the desired models, enter a name for your API key, and then click “Create”.  Figure 2: API Key creation interface  Verify the information and retrieve your newly generated API Key.  Figure 3: API Key successfully created 4. Configure Cursor IDE with FPT AI Marketplace API Steps to configure:  1. Open Cursor IDE → go to Cursor Settings → select Models.  2. Add Model:  Click Add model  Add the model (e.g., qwen_coder, deepseek_r1).  3. Enter API Key:  In the OpenAI API Key field, paste the API key you generated from FPT AI Marketplace.  4. Configure FPT AI URL:  Enable Override OpenAI Base URL  Enter the following URL: https://mkp-api.fptcloud.com  Figure 4: Configuring API Key and Base URL in Cursor IDE  5. Confirmation:  Click the Verify button.  If Verified Successfully appears, you are now ready to start using the model!  5. Using Code Generation Models in Cursor You can now:  Use the AI Assistant directly within the IDE to generate code.  Ask the AI to refactor, optimize, or explain your existing code.  Select the model you wish to use.  Figure 5: Using the Llama-3.3-70B-Instruction model from FPT AI Marketplace to refactor code  6. Monitor Token Usage To manage your usage and costs:  Go to My Usage on FPT AI Marketplace.  View the number of requests, input/output tokens, and total usage.  This allows you to see how many tokens you have used, helping you better control and manage your costs.  Conclusion  With just a few simple steps, you can harness the full power of the FPT AI Marketplace. You’ll be able to leverage advanced AI models at a cost-effective rate, accelerate your workflow with fast code generation, intelligent code reviews, performance optimization, and automated debugging. At the same time, you can easily monitor and manage your usage with clarity and transparency.     

FPT AI Factory has officially achieved AICPA SOC 2 Type I, SOC 2 Type II, and SOC 3 certifications, marking a significant milestone in our ongoing commitment to enterprise-grade security and global compliance. Particularly, the FPT AI Factory site in Vietnam attained SOC 2 Type II and SOC 3, while the FPT AI Factory site in Japan reached SOC 2 Type I. These achievements reaffirm FPT’s position as a trusted AI infrastructure provider for organizations and developers seeking to build and scale AI solutions with confidence. Understanding SOC Certifications: Setting the Global Standard for Trust The System and Organization Controls (SOC) standards were developed by the American Institute of Certified Public Accountants (AICPA) to evaluate how organizations manage data and ensure protection across key trust service principles: Each SOC certification provides a distinct level of assurance: SOC 2 Type I assesses whether security controls are suitably designed and implemented at a specific point in time. SOC 2 Type II evaluates the operational effectiveness of those controls over an extended period, typically from six to twelve months. SOC 3, designed for public distribution, offers a summarized version of SOC 2 findings, providing transparent assurance of FPT AI Factory’s commitment to global best practices in information security and privacy. By achieving all three certifications, FPT AI Factory demonstrates not only compliance but also maturity in operational excellence and a proactive approach to protecting client data. What Does This Mean for Our Clients? For customers deploying AI solutions at scale, data security and compliance are critical to success. Achieving SOC 2 and SOC 3 compliance means that FPT AI Factory’s systems, infrastructure, and internal processes have been rigorously evaluated by independent auditors for both design and effectiveness. This ensures that our clients, from global corporations to startups, can trust FPT AI Factory to handle their AI workloads, models, and datasets with the highest levels of protection. Key benefits for clients include: Verified data protection and privacy controls, aligned with global standards. Operational resilience for mission-critical AI and cloud workloads. Assurance for compliance with international regulations like GDPR, ISO 27001, and other frameworks. With SOC 2 and SOC 3 compliance, enterprises can now leverage these solutions knowing that every layer, from data management to model deployment, is protected by rigorous governance and independently verified controls. Building a Secure Foundation for AI Innovation Achieving SOC certifications is not the finish line, but a milestone in FPT AI Factory’s journey toward continuous improvement. We will continue to strengthen our internal governance, security frameworks, and audit processes to maintain the highest levels of reliability and transparency. This commitment reflects our broader mission: empowering organizations to innovate with AI securely and responsibly, building trust as the cornerstone of every intelligent system.

  We continue to advance the FPT AI Factory platform to improve scalability, performance, and operational efficiency. This release, as of November 17, 2025, introduces new features and optimizations designed to enhance the smoothness and efficiency of your workflows. FPT AI Studio Accelerate LLM workflows with new optimization techniques and gain real-time visibility through Grafana-integrated UI Logs. New Feature 1. Optimized LLM Performance Boost training speed and efficiency with new optimization techniques - liger_kernel, unsloth_gradient_checkpointing, and flash_attention_v2, reducing compute and memory costs for larger, faster workloads. 2. Enhanced Observability Gain deeper insights with new UI Logs integrated with Grafana, making it easier to monitor performance and troubleshoot in real time. AI Notebook Use and manage your workloads on AI Notebook more easily with the updated features. New Features 1. GPU Kernel Management Add a GPU resource management feature that allows customers to view the history of GPU kernel activation/deactivation. Enable or disable kernel flavors based on GPU resources to help customers connect with available GPU kernels. 2. Kernel flavor management by GPU availability so customers connect to suitable GPU kernels. 3. Improved long-term system stability and performance. Billing Experience flexible billing options and transparent cost control when using services and managing payment on FPT AI Factory. New Features Clearly define the services and products available to customers, including GPU Container, Model Fine-tuning, and more. Display the total plan amount (credit limit), which will automatically switch to on-demand pricing using Vouchers or Top-Up Credits once the plan expires. Allow customers to track their consumption via Billing → Credit History and view detailed usage under Billing → Billing Plan → View Details. Use Cases Set a custom price for all products, either for all PAYG customers or specific tenants. Let enterprise customers use postpaid Billing Plans with a defined spending limit. Allow customers to pay for their Billing Plan via offline bank transfer. 👉  Explore now: https://ai.fptcloud.com/undefined/billing Need help? Check our quick guide here or contact us with a click. Stay connected with us on LinkedIn & Facebook for the latest updates!

Today’s computer vision systems are highly effective at detecting what happens in physical environments: identifying objects, anomalies, or events. However, they still struggle to explain why those events matter, articulate fine-grained scene details, or reason about what could happen next. Agentic intelligence powered by vision language models (VLMs) can help bridge this gap, giving teams quick, easy access to key insights and analyses that connect text descriptors with spatial-temporal information and billions of visual data points captured by their systems every day. There are three practical ways organizations can upgrade their existing computer vision systems by integrating agentic AI capabilities: Apply dense captioning for searchable visual content. Augment system alerts with detailed context. Use AI reasoning to summarize information from complex scenarios and answer questions. Making Visual Content Searchable With Dense Captions Traditional video search tools built on convolutional neural networks (CNNs) often lack context and semantic depth. They are optimized for narrow tasks such as object detection but cannot describe scenes or convert vision into text. As a result, teams still spend significant time manually reviewing footage to extract insights. By embedding VLMs into existing applications, businesses can automatically produce highly detailed captions for both images and videos. These captions transform raw visual data into rich, searchable metadata, enabling flexible search beyond simple filenames or labels. This approach is already proving its value. For example, advanced inspection platforms have used VLM-powered understanding to transform millions of images into structured reports, dramatically improving accuracy and reducing manual effort. Systems enhanced with agentic AI have achieved up to 96% defect-detection accuracy, compared with roughly 24% using manual inspection, reducing downtime and improving overall quality control. For enterprises in manufacturing, transportation, and public services, dense captioning enables transparent, consistent insights essential for compliance, safety, and operational excellence. Augmenting Computer Vision System Alerts With VLM Reasoning CNN-based computer vision systems often generate binary detection alerts such as yes or no, and true or false. Without the deep reasoning powered by VLMs, these alerts may trigger false positives, overlook key details, or fail to provide context. This can lead to unnecessary operational costs, reduced trust in automation, and poor decision-making in safety-critical environments. Instead of replacing existing infrastructure, organizations can layer VLMs on top of current CV systems to create an intelligent review mechanism. When an incident is detected, the VLM adds context: clarifying where it happened, how it occurred, and why it matters. Smart-city applications have shown the power of this approach. For instance, Linker Vision uses VLMs to verify critical city alerts, such as traffic accidents, flooding, or falling poles and trees from storms. This reduces false positives and adds vital context to each event to improve real-time municipal response. Linker Vision’s architecture for agentic AI involves automating event analysis from over 50,000 diverse smart city camera streams to enable cross-department remediation, coordinating actions across teams like traffic control, utilities, and first responders when incidents occur. The ability to query across all camera streams simultaneously enables systems to quickly and automatically turn observations into insights and trigger recommendations for next best actions. Automatically Analyze Complex Scenarios With Agentic AI As organizations expand their sensor networks, spanning video, audio, text logs, and IoT devices, they need AI that can reason across all modalities, not just vision. This is possible by combining VLMs with reasoning models, large language models (LLMs), retrieval-augmented generation (RAG), computer vision, and speech transcription. A simple VLM integration is sufficient for verifying short clips, but standalone models are limited by the number of visual tokens they can process. This often results in shallow, surface-level answers. However, this approach is limited by how many visual tokens a single model can process at once, resulting in surface-level answers without context over longer time periods and external knowledge. In contrast, whole architectures built on agentic AI enable scalable, accurate processing of lengthy and multichannel video archives. This leads to deeper, more accurate, and more reliable insights that go beyond surface-level understanding. Agentic systems can be used for root-cause analysis or analysis of long inspection videos to generate reports with timestamped insights. Source: NVIDIA

One year ago, FPT took a bold step toward shaping the future of artificial intelligence, not just for Vietnam, but for the world. That vision became FPT AI Factory, a dynamic hub where innovation, computing power, and creativity converge. As we celebrate our first anniversary, let’s look back on a journey filled with breakthroughs, partnerships, and progress. From Vision to Innovation When FPT AI Factory was founded, the mission was clear: to build a world-class AI ecosystem that empowers organizations and individuals to design, train, and deploy their own AI models with unprecedented speed and flexibility. In just one year, that vision has transformed into tangible innovation. From advanced AI infrastructure to productized platforms, FPT AI Factory has become a key player in accelerating Vietnam’s AI ambition, while contributing to global progress in AI development. Built on the Strategic Partnership between FPT and NVIDIA The foundation of this journey lies in a strategic partnership between FPT and NVIDIA, uniting two forces with a shared purpose: democratizing access to high-performance computing and generative AI. Through NVIDIA’s cutting-edge technologies and FPT’s AI expertise, the collaboration has enabled the creation of scalable, high-performance AI infrastructure, which is the backbone for thousands of experiments, models, and innovations. Together, we are making it possible for developers, researchers, and enterprises to bring their ideas to life faster and more efficiently than ever before. Two AI Factories, One Global Mission In our first year, FPT AI Factory established two AI factories in Vietnam and Japan, serving as the twin engines of our innovation ecosystem. These AI factories are more than physical data centers. They are collaborative ecosystems designed to empower businesses and communities to co-create AI solutions. 11 New Products, Infinite Possibilities Innovation never stops at FPT AI Factory. In just one year, our team rolled out 11 new AI products, powering end-to-end AI innovation, fostering the mission to make AI creation accessible, efficient, and scalable, giving every researcher, developer, engineer, and business user the tools to Build Your Own AI. At the core of this growth are four powerful product lines: FPT AI Infrastructure: Providing enterprise-grade NVIDIA H100 & H200 GPUs for large-scale AI training and deployment, accelerating everything from foundational models to custom use cases. FPT AI Studio: Offering an all-in-one platform that enables users to create, fine-tune, and develop their own AI models. FPT AI Inference: Delivering optimized, cost-efficient inference pipelines to bring AI into real-world applications faster than ever. FPT AI Agents: Enabling instant AI adoption with an easy-to-use, multi-lingual platform to build and operate AI Agents. Make-in-Vietnam, for the World Over the past year, FPT AI Factory has demonstrated that Vietnam’s talent and technological capabilities can have a global impact. Our AI infrastructure and platforms now support initiatives that extend beyond borders, from research collaborations in Asia to AI deployments for global enterprises. Every milestone achieved at FPT AI Factory is a step toward redefining Vietnam’s role in the global AI landscape, not just as a participant, but as a leader and enabler. Looking Ahead As we celebrate one year of progress, we also look forward to new challenges, collaborations, and discoveries. FPT AI Factory will continue to champion open innovation and technological sovereignty, building an ecosystem where the global AI community can learn, experiment, and grow together. From vision to innovation, we are driving the momentum to Build Your Own AI. Here’s to many more! Explore FPT AI Factory: https://aifactory.fptcloud.com

FPT, a global ICT corporation and an NVIDIA Preferred Partner, introduced AI Notebook - a powerful, managed JupyterLab service that serves as a trusted coding companion for developers and researchers in day-to-day development. Built upon FPT AI Factory infrastructure, AI Notebook leverages NVIDIA accelerated computing and Jupyter Notebook open-source architecture to provide an elevated cloud-based coding workspace that allows AI engineers, developers, and researchers to prototype, experiment, and refine models — all faster, more securely, and collaboratively, with enterprise-grade reliability. [caption id="attachment_68075" align="aligncenter" width="1024"] A cloud-based platform for developers to accelerate AI research and development[/caption]   As organizations accelerate their adoption of AI, the demand for faster experimentation and more efficient model development continues to rise. Designed with developers in mind, AI Notebook eliminates Jupyter Notebook deployment hurdles to create a ready-to-use development environment and minimizes infrastructure overhead with optimal high-performance GPU options. This enables AI developers, data scientists, and students to shorten research and experimentation cycles, ultimately delivering results faster. Key benefits for the AI Notebook include: Accelerated experimentation and productivity: Provides a unified, pre-configured environment that gives developers a fast, intuitive experience to write and test code, explore data, build, and iterate on AI models interactively. That streamlines the workflow from early research to model training and fine-tuning, accelerating the journey from ideas to working models.  Performance at scale, payment on demand: Offers access to a range of NVIDIA H100 and NVIDIA H200 Tensor Core GPU configurations to match different stages of model development, delivering the performance needed to scale workloads seamlessly. A free starter setup is also available with no upfront cost, giving them sufficient capacity for basic experiments and evaluation before scaling to GPU acceleration. Flexible and transparent pay-as-you-go pricing with no hidden fees or data transfer charges ensures cost efficiency and the freedom to innovate. Enhanced collaboration and project management: Creates a collaborative space with advanced features that allow for running multiple projects in parallel, each workspace serving as a dedicated lab. Experiments and progress are centralized in one place, making it easy to compare results, reuse prior work, and move smoothly from research to production. Secure innovation: Built on NVIDIA AI infrastructure with enterprise-grade reliability, it ensures safe, compliant, and efficient AI development. Developers can innovate with confidence, knowing data and workloads are fully protected. Mr. Le Hong Viet, CEO of FPT Smart Cloud, FPT Corporation, emphasized, “Our vision is to empower every organization to build their own AI, tailored to their unique data, knowledge, and culture. With NVIDIA-accelerated FPT AI Factory and its next-generation GPUs, our platforms provide AI researchers, engineers, and developers with the tools to create, train, and scale models with enterprise-grade performance. By removing infrastructure barriers and optimizing costs, we make AI development more efficient, scalable, and practical — enabling organizations to innovate faster, smarter, and with greater independence.”   Availability Developers can sign up to explore AI Notebook while exploring other NVIDIA-accelerated services on FPT AI Factory. Visit https://ai.fptcloud.com/ to learn more and get started.

After the release of Claude Sonnet 4.5, considered a world-class model for programming and agentic use, Anthropic has introduced its newest small model: Claude Haiku 4.5. According to Anthropic, this model delivers better performance than Sonnet 4, while costing one-third as much and running at more than double the speed. Claude Haiku 4.5 is engineered for high-volume, low-latency, cost-sensitive deployments. If your workload involves long-running sequences, many calls to LLMs, or you need to spin up multiple agents in parallel, this is a major shift. Key technical highlights Claude Haiku 4.5 is described as a “small, fast model” in Anthropic’s classification. It sits below the “frontier” models but delivers near-frontier coding and reasoning performance at a much lower cost. On SWE-bench Verified (a real-world software engineering test using GitHub issues), Claude Haiku 4.5 scored ~73.3%. By comparison, Claude Sonnet 4.5 scored ~77.2%. Claude Haiku 4.5 supports both text and image inputs and is capable of extended reasoning, computer-use, and tool-assisted workflows. The model is available via Claude’s API at USD $1 per 1 million input tokens and $5 per 1 million output tokens. This is significantly lower than higher-tier models. In terms of safety and alignment, Anthropic assigns Haiku 4.5 under its AI Safety Level 2 (ASL-2) standard, which is a less restrictive classification than the ASL-3 assigned to the bigger models, and reports improved behaviour in alignment benchmarks. What this means for applications & users For developers, product teams, and businesses, Claude Haiku 4.5 opens up new possibilities: Cost-sensitive workflows: When you are running thousands or tens of thousands of model calls (e.g., customer service assistants, chatbots, embedded agents), the lower cost per token matters. Speed/latency-critical use cases: Claude Haiku 4.5 is faster, so it is well-suited for real-time interaction, multi-agent orchestration, or workflows where response speed is key. Scaling agents: If you architect a system with a top-tier model as the “brain” and multiple sub-agents handling sub-tasks, Claude Haiku 4.5 offers a faster, cheaper sub-agent tier without sacrificing too much in capability. Maintain high capability: Claude Haiku 4.5 offers near what was considered cutting-edge only months ago, along with more affordable pricing for many real-world coding, tool-use, and reasoning tasks. Flexibility in deployment: Claude Haiku 4.5 is available on Claude Code and Anthropic’s apps. Developers can access the model via API and on major cloud platforms (e.g., Amazon Bedrock, Google Cloud’s Vertex AI), making model adoption smoother. Conclusions The era when only the most expensive models could deliver top performance is changing. With Claude Haiku 4.5, Anthropic offers a compelling value proposition: remarkable performance, fast speed, and significantly lower cost. For organizations looking to embed AI agents, deploy at scale, or experiment with generative AI workflows, this model opens doors that were previously constrained by budget or latency. If you are working on AI-powered systems (chatbots, cloud agents, generative workflows), Claude Haiku 4.5 may well allow you to iterate faster, deploy more broadly, and keep your TCO (total cost of ownership) in check. Source: https://www.anthropic.com/news/claude-haiku-4-5

Artificial Intelligence (AI) has emerged as a strategic tool as the automotive industry pursues ambitious goals, such as improving operational efficiency, enhancing customer experiences, and prioritizing environmental sustainability.   1. The Significance of AI in the Automotive Industry With a suite of breakthrough features, AI is enhancing operational performance and optimizing production processes within the automotive sector. According to NVIDIA, European manufacturers, including BMW Group, Maserati, Mercedes-Benz and Schaeffler, are integrating AI into smart production lines, enabling real-time data monitoring and analysis to improve product quality, minimize errors, and increase precision at every stage.   Additionally, AI technologies also support long-term customer relationships by delivering personalized services such as virtual assistants, autonomous driving systems, and after-sales services.  2. Applications of AI in the Automotive Industry Manufacturing and Supply Chain Management  Artificial Intelligence serves as a critical enabler in automotive production and supply chain management. Companies like BYD have implemented AI on production lines, leveraging intelligent robots and predictive systems to optimize workflows, minimize errors, and increase productivity. These robots are capable of learning and self-adjusting while improving accuracy in tasks such as welding, assembly, and quality inspection.  In supply chain management, AI helps forecast market demand, optimize warehouse operations, and manage logistics efficiently. A notable example is Ford, which uses AI to analyze production and supply chain data, enhancing tracking capabilities, precise component transportation, and on-time delivery.   Furthermore, this technology also contributes to reducing carbon emissions and optimizing operational costs. These benefits are driving the automotive industry closer to a fully smart manufacturing model.   AI in Vehicle Features to Assist Drivers  To ensure safety and elevate the customer experience, leading automakers are deeply integrating AI into modern vehicles. One of the most common applications is Advanced Driver Assistance Systems (ADAS), which include automatic braking, lane departure warnings, adaptive cruise control, and driver drowsiness detection. These systems use artificial intelligence to analyze sensor and camera data, identifying risks and assisting drivers promptly, thereby reducing the likelihood of accidents.  Additionally, in modern vehicles, the Surround View Monitoring (SVM) systems harness artificial intelligence to provide a full 360-degree view of the surroundings. Combining information from cameras and sensors, these systems guide drivers through narrow spaces, facilitate accurate parking, and reduce the risk of collisions. By enhancing situational awareness, SVM contributes to both safety and comfort, providing drivers with a more intuitive and enjoyable experience behind the wheel.   After-Sales Services  Automakers are increasingly integrating artificial intelligence into after-sales services to enhance customer satisfaction and streamline maintenance processes. Intelligent systems can analyze large volumes of vehicle usage data to predict maintenance needs, schedule timely repairs, and suggest parts replacements before issues arise.  In Vietnam, VinFast has implemented AI-driven diagnostics and predictive maintenance tools to provide proactive support to its customers. Globally, Toyota leverages AI to optimize service scheduling, monitor vehicle health, and offer personalized recommendations, improving the overall ownership experience. 3. AI Factory: Providing Next-Gen AI Infrastructure for the Automotive Industry Robust AI infrastructure serves as a foundational platform, enabling companies to fully leverage the potential of artificial intelligence. With capabilities to process and analyze massive datasets, a strong infrastructure not only optimizes operational performance but also unlocks opportunities for groundbreaking AI innovations. This is particularly crucial in the automotive sector, where technological advancements and data-driven insights can drive significant progress in developing advanced solutions—from autonomous driving to optimized production and supply chain management. Therefore, AI Factory has become an indispensable pillar for the manufacturing industry.   In Europe, NVIDIA is building the world’s first AI factory that will support industrial AI workloads for European manufacturers. This Germany-based AI factory will feature 10,000 GPUs, including through NVIDIA DGX™ B200 systems and NVIDIA RTX PRO™ Servers, and enable Europe’s industrial leaders to accelerate every manufacturing application.  Furthermore, cloud providers in Japan, such as GMO Internet, Highreso, KDDI, Rutilea, SAKURA Internet, and FPT are leveraging NVIDIA-powered AI infrastructure to revolutionize industries including robotics, automotive, healthcare, and telecommunications.  In Asia, FPT launched FPT AI Factory in Japan and Vietnam, equipped with thousands of cutting-edge NVIDIA H100/H200 GPUs, delivering exceptional computing power. With this computational strength, businesses are allowed to drastically reduce research time while accelerating AI solution development and deployment by more than 1,000 times compared to traditional methods. This creates vast opportunities for turning ideas into reality and applying AI to enhance efficiency and innovation across all areas.