Every new instance in Frankfurt, Ashburn, and Singapore now provisions on gen5 NVMe by default. Existing gen4 volumes migrate automatically at your next reboot — no data movement required.
Our new PoP in Warsaw is now taking traffic. Peered at THINX Warsaw, transit via Arelion and GTT. Median RTT to FRA: 22ms.
Adds nodara_snapshot_schedule, drops the deprecated size attribute in favour of plan_id. See migration notes in the provider docs.
Balance forecasting, per-instance burn rate, and CSV export for invoices are all live in the dashboard. Also: we finally show which invoice a snapshot belongs to.
NVIDIA H200 with 141GB HBM3e is now bookable at $3.68/hr in IAD and FRA. Includes CUDA 12.6 preinstalled and a validated PyTorch 2.5 image.
We used to charge $0.005/GB over your plan cap with no ceiling. Now it caps at $30 per instance per month, so nobody wakes up to a five-figure bill.
Nodara's first African PoP. Peered at NAPAfrica JB1. Median RTT to LHR: 148ms.
Rewritten in Rust, single-file binary, zero-config auth via OAuth device flow. About 30x faster than the old Node CLI on cold start.
Six months of consistent performance let us bump the SLA one nine higher. Credit schedule updated in the same doc.