Ce que DigitalOcean a publié en avril 2026

Currents Research - February 2025

DigitalOcean’s February 2025 Currents report reveals 79% of growing tech businesses now use AI, up from 49% in 2023, but SMEs face barriers like GPU costs and lack of expertise. The report highlights …

fal-ai/wan/v2.2-a14b/text-to-video

Fal.ai released Wan 2.2-a14b, a major upgrade to its text-to-video model, adding image-to-video support, cinema-grade aesthetic controls, and improved motion realism. Pricing is now $0.08/second for 7…

Dedicated Inference

DigitalOcean introduced Dedicated Inference tools to its MCP server, enabling AI agents to run GPU-backed model workloads via the DigitalOcean API. The tools include create, update, get, delete, and l…

Introducing DigitalOcean AI-Native Cloud for Production AI Workloads

DigitalOcean introduced AI-Native Cloud, a full-stack system designed to simplify production AI workloads by reducing stack complexity and eliminating fragmented infrastructure. The platform includes …

How we built the most performant DeepSeek V3.2, MiniMax-M2.5 and Qwen 3.5 397B on DigitalOcean Serverless Inference

DigitalOcean announced general availability of optimized DeepSeek V3.2, MiniMax-M2.5, and Qwen 3.5 397B on its Serverless Inference platform, achieving #1 output speed across providers with metrics li…

News Details

DigitalOcean introduced its Inference Engine, a production-ready platform for AI inference workloads featuring Inference Router, Batch Inference, Serverless Inference, and Dedicated Inference. The sol…

DigitalOcean introduced the AI-Native Cloud, an end-to-end platform purpose-built for inference and agentic AI workloads. The five-layer stack integrates infrastructure, core cloud, inference, data, a

XiaomiMiMo/MiMo-V2.5-Prolike723FollowXiaomi MiMo3.87k

Xiaomi released MiMo-V2.5-Pro, an open-source Mixture-of-Experts model with 1.02T total parameters and 42B active parameters, featuring a 1M-token context window and hybrid attention architecture. It …

XiaomiMiMo/MiMo-V2.5like377FollowXiaomi MiMo3.87k

Xiaomi introduced MiMo-V2.5, a native omnimodal model supporting text, image, video, and audio with a unified architecture. Built on the MiMo-V2-Flash backbone, it features a hybrid attention design r…

Knowledge Bases

DigitalOcean introduced a standardized MCP endpoint for AI agents to search and retrieve content from knowledge bases via hybrid semantic and keyword search. The endpoint supports multiple protocol ve…

DigitalOcean Dedicated Inference: A Technical Deep Dive

DigitalOcean introduced Dedicated Inference, a managed LLM hosting service on dedicated GPUs with Kubernetes-native orchestration, designed for high-volume, sustained inference workloads. The service …

deepseek-ai/DeepSeek-V4-Prolike4.76kFollowDeepSeek135k

DeepSeek AI released the DeepSeek-V4 series, featuring two MoE models—DeepSeek-V4-Pro (1.6T params, 49B activated) and DeepSeek-V4-Flash (284B params, 13B activated)—both supporting a 1M-token context…

Beyond the Abyss Project Poseidon’s Quest for Zero-Downtime Reliability

DigitalOcean introduced Project Poseidon, an AI-driven system combining ML and GenAI to predict hypervisor crashes before they occur. The system uses tiered telemetry filtering and semantic log analys…

From Incident Counting to SLIs: How DigitalOcean Rethought Availability

DigitalOcean abandoned its incident-based availability metric in early 2025 after discovering it misaligned with customer experience, oscillating between 99.5% and 99.9% monthly. The new SLI-based fra…

Balaji Varadarajan

The DigitalOcean community profile page for Balaji Varadarajan highlights his role as a Staff Engineer. The page includes links to DigitalOcean’s documentation, startup and AI-native business resource…

The LLM Inference Trilemma: Throughput, Latency, Cost

DigitalOcean’s engineering blog explains the LLM inference trilemma—balancing throughput, latency, and cost—highlighting why scaling LLM inference is fundamentally different from traditional web servi…

Brett Snyder

The DigitalOcean community profile page for Brett Snyder, a Principal Engineer, highlights his role and contributions. Snyder has written 2 blog posts and has not yet created tutorials. The page promo…

Mastering the 600B+ Frontier: Optimizing Large Model Deployments on the Inference Cloud

DigitalOcean announced optimizations to its Spaces Object Storage and High Performance Managed NFS products to address the growing challenge of deploying large AI models (600B+ parameters, 700GB+ weig…

The Inference Cloud Memory Layer: A Technical Dive into DigitalOcean Managed Databases

DigitalOcean introduced a 'memory layer' for its Inference Cloud, positioning Managed Databases (PostgreSQL, MongoDB, Valkey, OpenSearch) as the foundational system of record for stateful AI applicati…

DigitalOcean announced that AI-native startups are migrating from hyperscalers to its Agentic Inference Cloud, citing 50% faster training cycles and 40% lower latency for production AI workloads. Star

Mohammad Ashar Khan

The page is a user profile for Mohammad Ashar Khan, a Senior Software Engineer at DigitalOcean, highlighting his minimal community contributions (1 blog post written, no tutorials or comments). The pa…

Load Balancing and Scaling LLM Serving

DigitalOcean’s engineering team details how prompt caching in LLM serving fundamentally changes load balancing requirements, as naive round-robin routing reduces cache hit rates linearly with fleet gr…

DigitalOcean Updates Documentation and Product Guidance for AI Agents and Modern Workloads

DigitalOcean has restructured its documentation to optimize AI agent access, introducing a full-text index (`llms.txt`), a structured JSON index (`llms-index.json`), and chunked documentation for prac…

Customers - Picap

Picap, a Latin American ride-sharing and logistics company, migrated from AWS and Heroku to DigitalOcean, reducing monthly infrastructure costs from $120,000 to $30,000 while improving performance. Th…

Shaoni Mukherjee

DigitalOcean has hired Shaoni Mukherjee as an AI Technical Writer, leveraging her six years of experience in data science and focus on AI, machine learning, and GPU computing. Her role involves creati…

Meghan Grady

The page is a user profile for Meghan Grady, who holds the title of Head of Marketing & Communications at DigitalOcean. The profile highlights her role and includes links to DigitalOcean’s community r…

Najmus Saqib

The DigitalOcean community profile page for user 'nsaqib' highlights their minimal engagement, with only one comment made and no tutorials or blog posts written. The page also promotes DigitalOcean’s …

Write for DOnations

DigitalOcean launched the 'Write for DOnations' program, a paid community writing initiative where authors create technical tutorials on open-source software, Linux systems, Docker, and security topic…

Andrew Dugan

DigitalOcean highlights Andrew Dugan, a Senior AI Technical Content Creator II with 8 years of experience in NLP and enterprise AI applications, as part of its community contributors. The company prom…

Building a Robust Documentation Agent with DigitalOcean Gradient AI Platform

DigitalOcean launched an AI-powered documentation assistant built on its Gradient AI Platform to help users find answers faster in its documentation. The agent allows users to ask questions in plain l…

pydo.dedicated_inferences

DigitalOcean introduced Dedicated Inferences, a scalable production-grade LLM hosting service, as part of its pydo v0.30.0 release. The service allows users to create, manage, and delete dedicated inf…

pydo.agent_inference

DigitalOcean introduced the Agent Inference API as part of its Gradient™ AI Agentic Cloud, enabling developers to integrate multi-agent workflows into AI applications. The API operates via a customer-…

pydo.inference

DigitalOcean introduced a new serverless inference API for AI agents, accessible via the pydo.inference module in version 0.30.0 of its Python SDK. The API enables access to serverless models for gene…

Making sure you're not a bot!

The Fedora Project has released its latest version, Fedora 40, introducing significant updates to its Linux distribution. Key changes include the default adoption of GNOME 46, which enhances desktop p…

App Platform Release Notes

DigitalOcean’s App Platform has introduced several updates since late 2024, including new buildpacks for Go, PHP, Python, Node.js, Ruby, Rust, .NET, Hugo, and Bun, along with general availability for …

Release notes

DigitalOcean has rolled out a series of incremental updates and strategic shifts across its cloud platform, App Platform, AI services, and security offerings, all aimed at enhancing performance, cost …

DigitalOcean has rolled out a series of incremental updates and strategic shifts across its cloud platform, primarily focused on enhancing performance, security, and AI/ML capabilities. The most signi

DigitalOcean has introduced several incremental updates and feature enhancements across its cloud platform, App Platform, AI services, and security offerings, all of which collectively improve perform

DigitalOcean has rolled out a series of incremental updates and strategic shifts across its cloud platform, AI services, and infrastructure offerings, with the most significant changes centered on AI

DigitalOcean has introduced several incremental updates and strategic improvements across its cloud platform, AI services, and infrastructure in recent months, with changes spanning product features,

DigitalOcean has introduced several incremental updates and new features across its cloud platform, AI services, and infrastructure in recent months, with the most significant changes affecting AI wor

DigitalOcean has introduced several incremental updates and strategic shifts across its cloud platform, primarily focused on AI/ML capabilities, Kubernetes enhancements, and security improvements. The

DigitalOcean has introduced several incremental updates and strategic shifts across its cloud platform, AI services, and infrastructure offerings, with changes spanning AI model availability, Kubernet

How to Manage Payment Methods on DigitalOcean

DigitalOcean updated its payment method management documentation on April 8, 2026, expanding accepted payment options and clarifying default payment behavior. The most significant change is the additi…

Bikram Gupta

The page is a user profile for Bikram Gupta on DigitalOcean’s community platform, highlighting his contributions such as writing 13 blog posts and visiting tutorials. It emphasizes opportunities for c…

Advanced Prompt Caching at Scale

The article introduces advanced prompt caching strategies to address a critical scalability challenge in large-scale AI inference deployments. While modern inference engines like vLLM, SGLang, and Ten…

zai-org/GLM-5.1like1.84kFollowZ.ai17.9k

Z.ai released GLM-5.1, a next-generation flagship model designed for agentic engineering tasks, outperforming its predecessor on SWE-Bench Pro and NL2Repo benchmarks. Unlike prior models that plateau …

10 Leading AI Cloud Providers for Developers in 2026

DigitalOcean’s February 2026 research found 46% of organizations deploying AI agents, driving demand for inference-focused cloud infrastructure. The article positions DigitalOcean’s Gradient AI Agenti…

High-performance business-ready block storage

DigitalOcean has enhanced its Block Storage product with NVMe-based volumes, offering low-latency, high-throughput storage for Droplets and Kubernetes clusters. The solution is designed for business-c…

Model Support Policy

DigitalOcean’s Gradient AI Platform has updated its model support policy to clarify the lifecycle of foundation and embedding models, including deprecation timelines and migration requirements. The pl…

Kubernetes at scale managed for you

DigitalOcean has enhanced its managed Kubernetes service, DigitalOcean Kubernetes (DOKS), with four new updates aimed at simplifying multi-cluster management while improving performance, reliability, …

Load balancing, simplified

DigitalOcean has introduced its Network Load Balancer as Generally Available (GA), expanding its load balancing capabilities with a new service that routes connections to Droplets and Containers withi…

Worry-free database hosting

DigitalOcean has enhanced its Managed Databases service with autoscaling for storage, allowing customers to automatically scale storage capacity without downtime or manual intervention. The offering s…

DigitalOcean Cloud Security Posture Management

DigitalOcean has launched Cloud Security Posture Management (CSPM), an agentless tool designed to scan DigitalOcean resources for misconfigurations without requiring third-party tools or agents. The s…

Supercharge your online presence with Cloudways managed hosting

Cloudways, a managed cloud hosting platform, introduces a streamlined solution for PHP developers, Laravel teams, and agencies by eliminating server management complexities. The platform enables one-c…

The Hidden Cost of Complex AI Platforms: Why Developer Experience Matters

The article highlights the hidden costs of complex AI platforms, emphasizing that developer experience (DX) is a critical but often overlooked factor in AI system deployment. While AI platforms offer …

Droplet Features

DigitalOcean has expanded its GPU Droplets offerings with new hardware configurations and specialized features for AI/ML workloads. The update introduces self-serve GPU Droplets with up to 10 Gbps pub…

The Glue Problem in Modern AI Development

The article highlights the 'glue problem' in AI development, where the complexity of integrating disparate cloud services—such as storage, compute, orchestration, networking, authentication, and infer…

Private Dropletspublic

DigitalOcean has introduced Private Droplets, a new type of Linux-based virtual machine (VM) that operates exclusively within a Virtual Private Cloud (VPC) without public internet connectivity. Unlike…

DigitalOcean has acquired Katanemo Labs, a leader in agentic AI infrastructure, to enhance its Agentic Inference Cloud platform. The acquisition brings Katanemo’s open-source project Plano, a data pla

The Agentic Era Demands a New Class of Infrastructure: Why DigitalOcean is Acquiring Katanemo Labs, Inc.

DigitalOcean has acquired Katanemo Labs, Inc., a leader in agentic AI infrastructure, to accelerate its strategic shift toward building the 'Agentic Inference Cloud.' The acquisition centers on Katane…

The simpler way to launch your apps

DigitalOcean has significantly enhanced its App Platform, a fully managed Platform-as-a-Service (PaaS), to simplify app deployment and scaling for developers. The platform now supports seamless deploy…

Run Advanced Reasoning on DigitalOcean with Arcee AI's Trinity Large-Thinking

DigitalOcean and Arcee AI have launched Trinity Large-Thinking, a frontier-class reasoning model, in Public Preview on DigitalOcean’s Agentic Inference Cloud. This integration allows developers to run…

Pricing discovered for Digitalocean

Pricing updated for Digitalocean: - New tier: Gradient™ AI Platform (Custom pricing) - New tier: Gradient™ AI GPU Droplets ($0.76/mo) - New tier: App Platform (Custom pricing) - New tier: Droplets (Cu…

Now Available: DigitalOcean Cloud Security Posture Management (CSPM)

DigitalOcean has launched Cloud Security Posture Management (CSPM), a new agentless security tool designed to help customers detect and remediate infrastructure risks without third-party tools or dedi…

How to Use Serverless Inference on DigitalOcean Gradient™ AI Inference Hub

DigitalOcean has introduced a new serverless inference capability within its Gradient™ AI Inference Hub, now in private preview. This feature allows users to send API requests directly to foundation m…