Running Spot GPUs on DOKS automates interruption handling with automatic cordoning, draining, and PodDisruptionBudget enforcement. Cluster autoscaling dynamically scales your GPU capacity based on workload demand. Kubernetes capabilities such as labels, taints, and node affinity allow you configure on-demand node pools as a fallback when Spot capacity is reclaimed. Plus, reclaim events surface as native Kubernetes events.

Try it today ->

GPT-6 Astra, OpenAI's newest and most capable model is now available through DigitalOcean Inference Engine. It delivers OpenAI's largest reported gains in agentic computer use and coding, alongside a reported substantial improvement in staying within task scope, while using markedly fewer output tokens per task than comparable frontier models, with no separate OpenAI account or contract required.

Claude Fable 5.1 is now available through DigitalOcean Serverless Inference, the Agent Platform, and the ADK. A point release on Claude Fable 5, Anthropic reports gains across agentic coding, long-running agentic workflows, and knowledge work, including a jump from 42.0% to 55.8% on Terminal-Bench 4.0, plus a 75% cut to cache-read pricing that lowers overall cost by roughly 25-45% depending on workload. Anthropic recommends it as a direct upgrade wherever Fable 5 is used today.

Access the model now ->

Memphis (MEM1) is now available as a DigitalOcean data center region. MEM1 expands DigitalOcean's footprint to 21 data centers across 11 global regions.

For current regional coverage across all products, see the availability matrix.

MEM1 security certifications are documented in our Security Reports and Certifications Center.

As with every DigitalOcean region, your compute, storage, networking, and inference sit on one platform and one bill.

Z.ai's GLM-5.3 is now available on DigitalOcean Inference Engine, a post-training-only upgrade to GLM-5.2 delivering a 50% gain on Z.ai's Code Bench and state-of-the-art CyberGym vulnerability-discovery results, with a 1M-token context window and tunable low/high/max reasoning effort.

GLM-5.3-Flash, Z.ai's first natively multimodal model in the GLM-5 series, is now available through DigitalOcean Inference Engine. A hybrid sparse-and-linear attention architecture makes its 1M-token context window substantially cheaper to serve, delivering GLM-5.2-class-or-better coding and agentic performance at roughly one-tenth the cost, under the MIT License.

Access the model now ->

v5 Droplets are now generally available in the MEM1, RIC1, ATL1, and MKC1 data center regions. Powered by 5th Gen AMD EPYC™ processors, v5 delivers up to 30% higher performance per-core compared to our previous generation Droplets, while giving you the flexibility to configure vCPU, memory, and storage independently and pay for only the resources you use. You can deploy v5 in both Shared and General Purpose configurations across standalone Droplets and Kubernetes node pools, alongside all existing Droplet plans.

Qwen3.8-Max, a 2.4 trillion-parameter mixture-of-experts model, is now available on DigitalOcean Inference Engine through Serverless Inference. This is an updated release in the Qwen3.8 family: on top of the long-context and agentic strengths of the earlier Qwen3.8 model, Qwen3.8-Max adds native image and video input, and a 1M-token context window built in.

DeepSeek-V4-Pro-0813 is a 1.6T-parameter, 49B-active MoE reasoning model now available through DigitalOcean Serverless Inference and Inference Router. It scores 53 on the Artificial Analysis Intelligence Index, well above the open-weights median of 27, and supports a 1M-token context window under an MIT license with no commercial-use restrictions.

Alibaba's Qwen3.8-2.4T-A95B, a 2.4T-parameter MoE model with a 1M-token context window is now available via DigitalOcean Serverless Inference.