TLDR
- Start with Google Cloud's own tools: FinOps hub, Recommender, budgets and anomaly detection. In 2026 Google added spend cap budgets for some AI and serverless services, GKE workload recommendations in FinOps hub, and CUD Analysis.
- Native tools cover Google Cloud only and mostly recommend rather than act. Outside CUD purchases and spend caps, you apply each change yourself.
- Pick by gap: commitments (ProsperOps, DoiT PerfectScale for Commitments, Usage.ai), GKE (CAST AI, IBM Kubecost), BigQuery (Rabbit, Sedai), allocation and reporting (DoiT Cloud Intelligence, Vantage, CloudZero, Ternary), and configuration and architectural waste fixed with approval across Google Cloud and the rest of your cloud, data and AI estate (PointFive).
- 2026 changed the map: spend-based CUDs moved to a discounted-price billing model, Vertex AI became Gemini Enterprise Agent Platform, and DoiT bought SELECT.
- PointFive OS finds Google Cloud waste such as Cloud Logging, Cloud Storage and BigQuery storage billing issues, traces it to its technical cause and helps engineers fix it with approval; TokenShift, the AI agent control plane for your employee workstations, completes the PointFive platform.
Where PointFive fits: PointFive is for teams that already use FinOps hub and Recommender and want the waste those tools do not flag found and fixed, with Google Cloud reviewed alongside the rest of their cloud, data and AI bill. PointFive OS runs 500+ deep detections across 40+ platforms, and Nubank saved $8.2M in 9 months with PointFive. See how Nubank automated 3,000+ DynamoDB fixes.
Disclosure: PointFive wrote this guide, and PointFive is listed first because it is our platform, not because of a ranking. Every competitor statement comes from that vendor's own website, documentation or press release, checked on October 27, 2026, and linked in the sources. Where a vendor publishes nothing, we say so.
How we evaluated
We compared each option on six dimensions, unweighted:
- Google Cloud depth: Compute Engine, GKE, BigQuery, Cloud SQL, Cloud Run, Cloud Storage, logging and AI services.
- Action: recommendations, tickets or executed changes, and the permissions required.
- Rate optimization: resource-based and spend-based committed use discounts (CUDs).
- Scope beyond Google Cloud: other clouds, Kubernetes, data and AI.
- Allocation: mapping spend to teams and products.
- Pricing model: as publicly stated.
The tools
1. PointFive (PointFive OS and TokenShift)
Category: Agentic FinOps platform (waste detection and remediation)
Best for: Enterprises running Google Cloud alongside other clouds, data platforms and AI that see their costs but struggle to get waste fixed.
PointFive OS, "The agentic FinOps platform for your cloud, data, and AI", runs 500+ deep detections (DeepWaste) and traces each finding to its technical cause. On Google Cloud it covers Compute Engine, GKE, Cloud SQL, BigQuery, Cloud Run, Cloud Storage and Gemini Enterprise Agent Platform (formerly Vertex AI), with detections for Cloud Logging configuration, Cloud Storage lifecycle, idle Agent Platform endpoints, overprovisioned Cloud SQL storage and agentless GKE workload sizing. For BigQuery it maps spend to dataset, table and job, finds unused tables and flags datasets on the wrong storage billing model; one customer cut BigQuery storage costs 82% in five days. Teams investigate in Chat with the Pointer agent or through an MCP server. TokenShift, the AI agent control plane for your employee workstations, works alongside it.
- Strengths: Configuration findings beyond rightsizing and idle resources. Remediation as IDE prompts or approved automation, at the approval level each customer chooses. Read-only discovery. 40+ platforms. Rated 4.9/5 on G2.
- Limitations: No self-serve tier or public price list. BigQuery slot, reservation and commitment optimization is listed as coming soon. Automated remediation needs customer-authorized write permissions, which security teams will review.
- Pricing: Annual contract, custom; contact sales. Listed on AWS Marketplace.
- Choose it if: Your problem is acting on Google Cloud waste, not seeing it. Nubank saved $8.2M in 9 months with PointFive; Dolby is also a customer.
2. Google Cloud native tools (as of October 2026)
Category: Built-in cost analysis, recommendations, discounts and controls
Best for: Every Google Cloud customer, as the baseline.
- Recommender (Active Assist): cost recommendations for idle VMs, disks, IPs and GKE clusters, Cloud SQL idle and overprovisioned instances, CUDs, BigQuery edition slots and partitioning, Cloud Run CPU allocation and idle reservations. You apply them.
- FinOps hub: realized and potential savings, a FinOps score and Gemini Cloud Assist insights in one dashboard. GKE workload recommendations arrived April 20, 2026.
- Budgets and anomaly detection: anomaly detection went GA on October 30, 2025, with early AI workload signals added July 24, 2026. Spend cap budgets (GA October 13, 2026) pause new usage of Gemini API, Gemini Enterprise Agent Platform, Cloud Run or Cloud Run functions in one project when estimated spend passes the cap.
- Committed use discounts: resource-based CUDs, and spend-based CUDs such as Compute flexible CUDs (Compute Engine, GKE, Cloud Run) and CUDs for Cloud SQL, BigQuery, Spanner and more. Under the new spend-based model, usage bills at the discounted price instead of list price minus a credit, and commitments are sized in discounted spend; Google says total costs do not increase. Flexible Savings Plans commit to a monthly minimum spend across eligible products. CUD Analysis (GA June 1, 2026) shows both CUD types in one view.
- Sustained use discounts: automatic, up to 30% for N1 and up to 20% for N2, N2D and C2. E2 is not listed.
- Gemini Cloud Assist: explains cost changes on the Cloud Hub Optimization page (Preview); it does not change resources.
- GKE: cost allocation labels billing data by namespace and workload, based on requests, not usage. Workload recommendations flag workloads under 50% utilization for 90% of 15 days; you apply them.
- BigQuery: on-demand or Standard, Enterprise and Enterprise Plus editions with autoscaling; Enterprise tiers offer 20% (1-year) or 40% (3-year) commitments.
What native tools leave to you:
- One cloud. Nothing here sees AWS, Azure, Snowflake or Databricks.
- Implementation. Beyond CUD purchases and spend caps, you validate each recommendation, find the owner, make the change and confirm savings.
- Detection depth. Configuration waste, such as Cloud Logging routing or BigQuery storage billing models, needs investigation.
- Strengths: First-party data, no procurement. Spend caps are real enforcement for AI and serverless spend.
- Limitations: As above. Spend caps cover one project and one service each.
- Pricing: Budgets and anomaly detection free; Pub/Sub notifications bill at Pub/Sub rates. BigQuery billing export incurs BigQuery charges.
- Choose it if: Always, and first.
3. ProsperOps (as of October 2026)
Category: Autonomous commitment management
Best for: Teams that want Google Cloud CUDs bought and managed hands-off.
ProsperOps says "Maximize Google Cloud savings with minimal commitment risk". It buys and manages resource-based CUDs and spend-based Flexible CUDs, choosing the mix from billing data, across Compute Engine, GKE, Cloud Run, Cloud SQL, Cloud Composer, Dataproc and Vertex AI. Flexera acquired ProsperOps in January 2026. It is listed on Google Cloud Marketplace.
- Strengths: Fee tied to savings. Free savings analysis. AWS and Azure too.
- Limitations: Rate optimization only on Google Cloud. BigQuery is not on the managed service list.
- Pricing: A percentage of savings; rates not published.
- Choose it if: CUD coverage is your largest untapped lever.
4. DoiT (as of October 2026)
Category: Allocation and reporting, plus Kubernetes, commitment and data-platform automation
Best for: Teams that want analytics and automation from one Google Cloud partner.
DoiT's headline is "Attribute every dollar. Even the AI ones." Cloud Intelligence allocates spend and runs 800+ Insights checks. DoiT acquired PerfectScale (Kubernetes) in February 2025 and SELECT in January 2026. PerfectScale for Commitments (GA May 21, 2026) manages Google Cloud CUDs and AWS Savings Plans, autonomously or with approval. SELECT opened BigQuery automation early access on April 16, 2026 (storage billing, reservations, query routing), with GA "planned for Q3".
- Strengths: Allocation, Kubernetes and commitments under one vendor. Free to start.
- Limitations: Supported CUD types not published. BigQuery automation was in early access when last announced.
- Pricing: Cloud Intelligence prices not published; PerfectScale for Commitments is performance-based, "customers pay only when the platform delivers realized savings".
- Choose it if: You want one Google Cloud partner across analytics and automation.
5. CAST AI (as of October 2026)
Category: Kubernetes automation
Best for: Large GKE estates.
CAST AI says "Kubernetes workloads, cloud and AI infrastructure, tuned together." It automates workload rightsizing, node autoscaling, Spot placement and bin packing, and tunes IOPS and throughput on supported GKE disks. In January 2026 it launched OMNI Compute and announced a valuation above $1 billion.
- Strengths: Continuous automation on GKE, EKS, AKS and more.
- Limitations: Kubernetes first. Needs permission to act on clusters. No commitment management mentioned.
- Pricing: Custom quote; 30-day free trial.
- Choose it if: GKE compute is a large share of your bill.
6. IBM Kubecost (as of October 2026)
Category: Kubernetes cost visibility and allocation
Best for: Platform teams that need GKE cost by namespace and workload.
"IBM Kubecost helps teams continuously reduce the cost of operating Kubernetes." It supports GKE, EKS, AKS and on-premises clusters, and can turn recommendations into actions such as automated request sizing or namespace turndown.
- Strengths: Free Foundations tier. Self-hosted or SaaS.
- Limitations: Kubernetes only. Foundations caps at 250 cores and 15-day retention.
- Pricing: Foundations free; Enterprise prices not published.
- Choose it if: GKE visibility is your main gap.
7. Sedai (as of October 2026)
Category: Autonomous cloud optimization
Best for: Teams comfortable with continuous automated tuning, including BigQuery and GKE.
Sedai says it "safely lowers your cloud costs, on autopilot." On Google Cloud it right-sizes GKE and Compute Engine, adjusts Cloud Storage tiers and Persistent Disk types, and tunes Dataflow, BigQuery and Cloud SQL to query and pipeline behavior. Modes run from Datapilot (recommendations) to Copilot (approve each action) to Autopilot, with rollback in all of them.
- Strengths: Broad Google Cloud service list, from GKE to BigQuery. Also AWS, Azure and others.
- Limitations: Autonomous modes need write permissions. Its Google Cloud page does not detail BigQuery actions.
- Pricing: 1 Sedai Billing Unit per vCPU managed; prices not published.
- Choose it if: You want hands-off tuning across GKE, Compute Engine and data services.
8. Rabbit (as of October 2026)
Category: BigQuery optimization
Best for: BigQuery-heavy teams that want slot and storage decisions automated.
Rabbit calls itself "The autonomous engine for BigQuery". It runs optimizers for slots, pricing, storage, reservations, caching and SQL, and proposes SQL rewrites as pull requests. Automation is opt-in per optimizer. It is a Google Cloud Partner on Google Cloud Marketplace.
- Strengths: Focused on BigQuery depth.
- Limitations: BigQuery only. Commitment automation is not described in detail.
- Pricing: Not published; free trial.
- Choose it if: BigQuery is your largest Google Cloud line item.
9. Usage.ai (as of October 2026)
Category: Commitment automation
Best for: Teams that want CUDs bought with downside protection.
Usage.ai describes "Automated Cloud Cost Optimization with Cashback Protection". On Google Cloud it buys CUDs for Compute Engine, GKE and Cloud SQL, and pays cash back on eligible unused value under its Flex Insured Commitments. It also covers AWS and Azure.
- Strengths: No subscription; fee only on new savings.
- Limitations: Rate optimization only. Supported CUD types are not listed on the homepage.
- Pricing: A share of new savings; "If Usage.ai generates no new savings, you pay nothing."
- Choose it if: Commitment risk is what stops you from buying.
10. Vantage (as of October 2026)
Category: Cost visibility, allocation and recommendations
Best for: Engineering-led teams that want self-serve reporting.
Vantage's headline is "Understand every dollar from Agents to AWS". On Google Cloud it offers Compute Engine rightsizing, idle GKE cluster and compute commitment recommendations, plus commitment reports for CUD coverage and effective savings rate.
- Strengths: Free tier and public prices. Multi-cloud.
- Limitations: Autopilot commitment buying is AWS only. Google Cloud recommendations are advisory.
- Pricing: Free (up to $2,500 tracked spend), $30 and $200 per month, Enterprise custom.
- Choose it if: You want to start small without a sales cycle.
11. CloudZero (as of October 2026)
Category: Cost allocation and unit economics, positioned around AI ROI
Best for: SaaS and AI companies that need cost per customer or feature.
"CloudZero is the AI ROI Company." It ingests Google Cloud alongside other clouds and AI providers for allocation, unit costs and anomaly detection. AI Signals (September 2026) attributes AI spend to people and teams.
- Strengths: Unit cost metrics across providers.
- Limitations: No Google Cloud specific optimization features described.
- Pricing: Custom quote for "a single subscription with all capabilities included".
- Choose it if: Finance needs unit costs.
12. Ternary (as of October 2026)
Category: Technology spend allocation and reporting
Best for: Large finance teams that want Google Cloud in a CFO ledger.
Ternary now leads with "Technology Investment Intelligence for the CFO", mapping AI, SaaS and cloud spend to cost centers and GL codes. It ingests Google Cloud billing, metrics and CUD data, breaks GKE cost down to pod, and can run self-hosted in your own Google Cloud projects.
- Strengths: Deep Google Cloud billing ingestion. Self-hosted option.
- Limitations: Its homepage targets companies with $500M+ revenue. Recommendations and automated actions are not described.
- Pricing: Not published.
- Choose it if: Finance needs Google Cloud mapped to the general ledger.
Side-by-side comparison
| Tool | Primary job on Google Cloud | Beyond Google Cloud | Takes action | Pricing model |
|---|---|---|---|---|
| PointFive (PointFive OS + TokenShift) | Find and fix deep waste | 40+ platforms | Yes, with approval | Custom |
| Google Cloud native tools | Recommendations, discounts, caps | No | CUD purchases, spend caps | Free |
| ProsperOps | CUDs | AWS, Azure | Yes | Share of savings |
| DoiT | Allocation, GKE, CUDs, BigQuery | Multi-cloud | Yes | Not published, performance-based |
| CAST AI | GKE automation | Kubernetes anywhere | Yes | Custom |
| IBM Kubecost | GKE visibility | Kubernetes anywhere | Some automations | Free tier, custom |
| Sedai | Autonomous tuning, GKE to BigQuery | Multi-cloud | Yes | Per vCPU, not published |
| Rabbit | BigQuery | No | Yes, opt-in | Not published |
| Usage.ai | CUDs | AWS, Azure | Yes | Share of savings |
| Vantage | Visibility, allocation | Multi-cloud | Not on Google Cloud | Public tiers |
| CloudZero | Unit economics, AI ROI | Multi-cloud, AI | Not documented | Custom |
| Ternary | Finance ledger | Multi-cloud, SaaS, AI | Not documented | Not published |
"Not documented" means we did not find it on the vendor's own pages, not that the capability is absent.
How to choose
- CUD coverage is low: ProsperOps, Usage.ai or DoiT PerfectScale for Commitments. Remove waste first so you do not commit to it.
- Most compute is GKE: CAST AI or Sedai for automation; IBM Kubecost for visibility.
- BigQuery is the fastest-growing line: Rabbit or Sedai for automated tuning; PointFive for storage billing models and unused tables.
- You know where the money goes, but nothing gets fixed: PointFive for configuration waste and approved remediation.
- Nobody owns spend: DoiT, Vantage, CloudZero or Ternary.
- AI spend can run away: set spend cap budgets on Gemini API and Agent Platform projects first.
Choose by Google Cloud spend level
- Under $50K per month: Native tools first, plus a share-of-savings commitment tool if nobody manages CUDs. A custom-priced platform, including PointFive, is usually more than you need unless data or AI spend is growing fast.
- $50K to $1M per month: Automate CUDs, add GKE or BigQuery tooling where those lines are large, and decide whether your gap is allocation or remediation. Here unfixed waste usually outgrows native recommendations, and PointFive fits.
- Over $1M per month: Most companies combine a commitment engine, an allocation system and a remediation platform such as PointFive. Budget time for security review of write permissions, and measure realized savings, not estimates.
FAQ
What is the best tool to reduce Google Cloud costs?
There is no single best tool. Start with FinOps hub, Recommender and the free budgets and anomaly detection, then match the gap: ProsperOps or Usage.ai for CUDs, CAST AI or IBM Kubecost for GKE, Rabbit or Sedai for BigQuery, DoiT, Vantage or CloudZero for allocation, and PointFive to find and fix deeper waste across Google Cloud and the rest of your estate.
Are Google Cloud's native cost tools enough?
At modest, Google Cloud only spend, often yes. Beyond that, they stay inside Google Cloud and mostly stop at recommendations.
What changed in Google Cloud cost management in 2026?
Spend-based CUDs moved to a discounted-price billing model. Google added spend cap budgets, GKE workload recommendations in FinOps hub, CUD Analysis, a FOCUS billing export (Preview) and early AI anomaly signals. Vertex AI became Gemini Enterprise Agent Platform on April 22, 2026.
Do GCP cost tools change resources automatically?
Some do. Spend caps pause eligible services, CAST AI, Sedai and Rabbit change resources or settings, and ProsperOps, Usage.ai and DoiT buy CUDs. PointFive discovery is read-only; remediation runs under customer-authorized permissions.
How is PointFive different from Recommender and FinOps hub?
Recommender produces resource-level recommendations such as idle VMs and Cloud SQL rightsizing, and FinOps hub tracks them. PointFive investigates configuration and architectural waste, such as Cloud Logging bucket design, and helps engineers fix it with approval.
Methodology and sources
We reviewed Google's and each vendor's own website, pricing pages, documentation and press releases on October 27, 2026, and did not use third-party sites for competitor claims. If something is out of date, tell us at info@pointfive.co and we will correct it.
- Google Cloud: Recommender cost recommenders, Recommender pricing, FinOps hub, Cloud Billing release notes, cost anomalies, budgets, spend cap budgets, Cloud Billing pricing, new spend-based CUD model, Flexible Savings Plans, sustained use discounts, Gemini Cloud Assist cost optimization, GKE cost allocation, GKE workload recommendations, BigQuery editions, Gemini Enterprise Agent Platform launch
- ProsperOps: Google Cloud product page, pricing, Flexera acquisition, Google Cloud Marketplace listing
- DoiT: homepage, PerfectScale acquisition, SELECT acquisition, PerfectScale for Commitments, SELECT BigQuery early access
- CAST AI: homepage, pricing, OMNI Compute and valuation
- IBM Kubecost: product page
- Sedai: homepage, GCP optimization
- Rabbit: homepage
- Usage.ai: homepage
- Vantage: homepage, pricing, cost recommendations docs, GCP commitment reports
- CloudZero: homepage, pricing, AI Signals
- Ternary: homepage, Google Cloud integration docs