Skip to content

Compute future reservations vs Compute on demand reservations

Compute future reservations is an open-source agent skill for cloud and infrastructure. Compute on demand reservations is an open-source agent skill for cloud and infrastructure.

  • Compute future reservations is best for check whether 64 a3 gpu vms can be reserved for a two-week training run next month.
  • Compute on demand reservations is best for check whether reserved vms in a zone still have free capacity.
What it is
Compute future reservations is an open-source agent skill for cloud and infrastructure.
Compute on demand reservations is an open-source agent skill for cloud and infrastructure.
Best for
Check whether 64 A3 GPU VMs can be reserved for a two-week training run next month
Check whether reserved VMs in a zone still have free capacity
Kind
Skill
Skill
Pricing
Open source
Open source
Platforms
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf, Goose
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, Windsurf, Goose
GitHub stars
24.4K
24.4K
Last commit
Nov 2026
Nov 2026
License
Apache-2.0
Apache-2.0
Open source
Good for
Check whether 64 A3 GPU VMs can be reserved for a two-week training run next month; Find a zone with capacity for a TPU v6e slice starting in a given date range; Compare a3-highgpu-8g and a3-megagpu-8g availability for a planned training job
Check whether reserved VMs in a zone still have free capacity; Find out why a new VM is not consuming my reservation; Share a reservation with another project in the same organization

What Compute future reservations does

  • Calendar-mode feasibility queries

    Runs gcloud advice queries for GPU, H4D and TPU blocks across a start-time window and duration range.

  • Reservation request drafting

    Produces the gcloud future-reservations create command for a confirmed window, pending human approval.

  • Topology and chip validation

    Checks instance counts and TPU chip counts against the valid values for each accelerator version.

  • Consumption guidance

    Shows how to launch VMs or TPU capacity against the auto-created reservation at start time.

  • Fallback planning

    Suggests alternate regions, wider start windows, shorter durations or other hardware generations when capacity is unavailable.

  • Cancellation and billing rules

    Explains cancellation cutoffs, deletion preconditions, IAM roles and the billing exposure of approved requests.

What Compute on demand reservations does

  • Reservation status and capacity check

    Reads reservation status, assured, in-use and requested counts to calculate available reserved capacity.

  • Targeting policy verification

    Explains automatic versus targeted consumption and checks that VM properties match the reservation.

  • Shared reservation auditing

    Covers share settings across host and consumer projects and where shared reservations are visible.

  • Billing and quota warnings

    Flags immediate billing and quota use when reservations are created and the double-billing risk of mismatched VMs.

  • Guarded lifecycle commands

    Provides gcloud templates to create, resize, share and delete reservations, gated on human approval.

Compute future reservations vs Compute on demand reservations: questions

Is Compute future reservations or Compute on demand reservations cheaper?

Both are open source. Compare current plans on each official site.

What is the difference between Compute future reservations and Compute on demand reservations?

Compute future reservations is an open-source agent skill for cloud and infrastructure. Compute on demand reservations is an open-source agent skill for cloud and infrastructure.

Keep comparing