Skip to main content

2025 Releases

Releases marked with "Please update" include important client-side changes — update by running the Windows installer or the Linux/Mac installation scripts.

v0.52.1 (2025-12-21)​

Minor release, no update necessary.

Rift Server​

  • Standardized principal resolution (user or team) across VM rental, reservations, transactions, payment intents, and service charge APIs. All endpoints now obey a unified authentication scheme: user JWT token, user API key, team API key, or user authenticated via JWT token making changes on behalf of the team.
  • Filtered out inactive instances when listing networks.

v0.52.0 (2025-12-18)​

Please update clients.

Rift Server​

  • Added Swagger UI at api.cloudrift.ai/swagger-ui in addition to RapiDoc, resolving issues with some OpenAPI 3.1 semantics.
  • Added missing schemas for API documentation, including the missing type for /api/v1/instances/rent.
  • Added POST /api/v1/network/delete to delete networks.
  • Added POST /api/v1/network/update to update network team access. Set team_id to null to make a network public, or specify a team ID to restrict access. Supports team assignment by ID (ById) or by name (ByName).
  • Extended /api/v1/instances/list response for VM instances: virtual_machines[].state now exposes the hypervisor VM state (e.g., Running, Shutoff).
  • Enhanced POST /api/v1/network/list:
    • Full details for admin/provider: gateway, netmask, iface, server_address, team_id, IP ranges, executor info (id, ip, owner_email).
    • Minimal info for regular users: network names only (previous behavior).
    • Replaces the old list-ranges endpoint.
    • Fixed visibility: providers/admins now see team-scoped networks in datacenters they own.
    • New selectors: All, ByTeamId, ByProvider.
  • Extended POST /api/v1/network/add to support specifying a team by name via team_name.
  • L40S GPUs are now available for rental.

v0.51.4 (2025-12-16)​

Minor release, no update necessary.

Rift Server​

  • Added POST /api/v1/instances/start and POST /api/v1/instances/stop endpoints for VM lifecycle management.
  • Providers and admins can start/stop VMs on their nodes; users can manage their own instances.
  • Providers and admins can also use POST /api/v1/instances/terminate and POST /api/v1/instances/reset for instances on their nodes.

Rift CLI​

  • Added rift virt start and rift virt stop commands.

v0.51.3 (2025-12-12)​

Minor release, no update necessary.

Rift Server​

  • Added API to add and remove recipes, groups, and tags (for server owners only).
  • Extended /api/v1/instances/list with a new ByNodeId selector that allows providers and admins to list all instances on a specific node.

v0.51.2 (2025-12-08)​

Minor release, no update necessary.

Rift Server​

  • Allow admin users to manage networks in all data centers.

v0.51.1 (2025-12-05)​

Minor release, no update necessary.

Rift Server​

  • Extended /api/v1/capabilities to include provider_name and is_admin.
  • Store information about invalid GPUs in the database.
  • Skip invalid GPUs when checking resources for availability.
  • Added a new command to synchronize VM images.
  • Extended /api/v1/instances/list to include live memory information of virtual machines via Prometheus.

Rift Desktop​

  • Report GPU validity status to the server.
  • Syncing VM images when requested by the server.
  • Collection of VM metrics via pull-based HTTP call to OpenTelemetry, with scraping of metrics by Prometheus.

Rift Library​

  • Added OpenTelemetry metrics layer for exporting metrics. Exports metrics on /metrics endpoint to allow Prometheus scraping.

v0.51.0 (2025-11-27)​

Please update clients.

Rift Server​

  • Extended /api/v1/instance-types/list to include per-datacenter provider name, logo URL, and country code in the response.
  • Aligned auth HTTP statuses: 401 for authentication failures (e.g., invalid/expired token), 403 for authorization denials (teams/providers/API keys). Impacts: /api/v1/nodes/list, /api/v1/network/*, /api/v1/reservations/*, /api/v1/instances/*, /api/v2/payment_intent/create, /api/v1/services/*.
  • Introduced persistent volume storage. Volumes provide persistent data storage independent of instance lifecycle, allowing you to preserve data when instances are terminated and reuse storage across multiple instances.

Rift Desktop​

  • Support volume operations for VMs and Docker containers.

v0.50.1 (2025-11-21)​

Please update clients.

Rift Server​

  • Fixed a bug where reservations were not detached when an instance failed to initialize. Reservations are now unbound when the instance is deactivated.
  • Added VM settings per provider and datacenter, allowing customized configurations for VMs based on their deployment context.
  • Added a way to change server currency, used in email templates instead of hardcoded "USD".

Rift Desktop​

  • Keep domains if configured to do so, even after VM destruction, preserving VM state for future use.
  • Add additional disks to VMs based on their configuration. Allows the use of persistent disks stored on network storage.

v0.49.5 (2025-10-15)​

Minor release, no update necessary.

Rift Server​

  • Added instance name and cluster name when renting an instance.
  • Changed node allocation strategy: when renting, choose the node with fewest available GPUs to fulfill the request. This improves GPU resource utilization across nodes.
  • Added signup bonus ($1) for first-time users. Password users receive it after email verification; OAuth users receive it on first-time registration.

v0.49.4 (2025-10-02)​

Minor release, no update necessary.

Rift Server​

  • Added support for configuring ports to be exposed for virtual machines when renting on a datacenter with port tunneling.

Rift Desktop​

  • Support switching between NVIDIA and VFIO drivers at runtime to support container and VM mode on one machine.

v0.49.3 (2025-09-19)​

Minor release, no update necessary.

Rift Server​

  • Added POST /api/v1/reservations/create to attach a reservation to a running instance.
  • Added POST /api/v1/reservations/update to update the duration of a running reservation.
  • Reservation conversion: You can now convert running instances into reserved ones to save on costs.
  • Added support for configuring FRP tunnels on providers with limited public IPs. Each user can reach their individual instance under the same IP but with dynamically allocated ports (e.g., 8021 → 443 and 8573 → 3000).
  • Added ip_availability_per_dc field to InstanceVariantInfo in /api/v1/instance-types/list to provide per-datacenter public IP availability information.
  • Added more information to /api/v1/nodes/list endpoint: IP address, hostname, and status.
  • Free LLM model swap: The free inference model has been changed from meta-llama/Meta-Llama-3.1-70B-Instruct-FP8 to Qwen/Qwen3-Next-80B-A3B-Thinking.

Rift Desktop​

  • Added support for executing FRP commands on nodes running frpc.

Rift CLI​

  • Added rift gpu list command to list GPU devices with PCI information.

v0.49.2 (2025-09-04)​

Minor release, no update necessary.

Rift Server​

  • Moved /api/v1/providers/nodes/list to /api/v1/nodes/list. Admins may use an optional provider selector; providers see only their own nodes; others receive 401.
  • Added POST /api/v1/capabilities/list returning available features (e.g., nodes.read) for a user.
  • Support setting a credit limit on accounts for invoicing.
  • Added API endpoints to add, update, and delete instance types and variants, as well as set an instance type for a node.