Ce que Digitalocean a publié en juin 2026

Model Evaluations: Prove Your Routing Policy Actually Works

DigitalOcean introduced Model Evaluations in Public Preview, enabling users to test and compare inference models, routing strategies, cost, latency, and output quality before deploying to production. …

Now Available: NVIDIA Nemotron 3 Ultra

The Team Behind Deploy: Shipping AI, the DigitalOcean Way

DigitalOcean unveiled its AI-Native Cloud, a five-layer stack for AI-native companies, featuring over 15 product launches including Inference Router, Dedicated Inference, Managed Weaviate, Knowledge B…

Powering the Inference Era: Inside the DigitalOcean Data & Learning Layer

DigitalOcean introduced a unified Data & Learning layer to unify structured and vector data for AI-native applications. Key updates include PostgreSQL and MySQL Advanced Edition (Public Preview) for h…

Now Available: Claude Opus 4.8

Open by Design: How NVIDIA and DigitalOcean Are Building the Stack for the Always-On Agentic Era

DigitalOcean and NVIDIA highlighted the importance of open-source AI models and infrastructure for building agentic systems in a Deploy 2026 session. They emphasized model flexibility, evaluation stan…

DigitalOcean Serverless Inference: A Deep Dive

DigitalOcean introduced Serverless Inference, a fully managed API-first platform supporting 30+ foundation models across text, code, vision, image, video, and speech. It offers single-endpoint access,…

The Inference Tax: How Prefix-Aware Routing Eliminates the Hidden Cost of LLMs at Scale

DigitalOcean unveiled a prefix-aware routing and caching mechanism to eliminate redundant LLM inference costs, targeting the 'prefill tax' where identical system prompts and shared contexts are recomp…