Now available: @Alibaba_Qwen 3.8-2.4T-A95B from @alibaba_cloud on DigitalOcean Serverless Inference via NVIDIA HGX™ B300 GPUs. 🤖 1M context, built for long-horizon coding. 🔗 do.co/45VBAyZ
One API, usage-based pricing, no infra to manage.
DigitalOcean
@digitalocean
- Latest-gen GPUs, without long term commitments. ☁️ Spot GPU Droplets are now in Public Preview on DigitalOcean. •NVIDIA HGX™ B300, AMD Instinct™ MI355X and MI350X, available now •Built for fault-tolerant work: batch training, batch inference, rendering •Same platform as
- DeepSeek-V4-Flash-0731 is live on DigitalOcean. 🆕 do.co/4xsz0fI A full post-training pass for agentic coding: Terminal Bench 2.1 up 61.8 to 82.7, now beating @deepseek_ai's own V4-Pro, at 13B active params. Serverless and Dedicated Inference, predictable pricing, no
- Struggling to find available GPUs? We've got you. Now available on-demand: NVIDIA HGX™️ H100 and AMD Instinct™️ MI325X. Scale up AI workloads without the complexity: • Enterprise-grade reliability • HIPAA-eligible and SOC-2 compliant • Integrated with the DigitalOcean
- Independent validation is the best kind. @OmerShlomovits got the full @Kimi_Moonshot K3 model running on a single @AMD 8xMI325X node on DigitalOcean, and wrote up every fix it took to get there. 🌩️ Worth a read if you're evaluating inference infra for large MoE models. 👇Today we added @AMD MI325X as a new serving backend for @zroai_. Kimi K3 is the first model running on it, powered by @sgl_project. We believe this is the first production support for the official Kimi K3 weights on CDNA3. With 256 GB of HBM per GPU, the full model fits on a
