Operator · Mountain View, CA
GMI Cloud
US/Taiwan GPU cloud selling dedicated NVIDIA H100, H200, B200 and GB200 capacity (containers, bare metal, managed Kubernetes clusters) alongside a serverless model-inference API.
GPU types4with a current offer
Current offers1214 readings on file · last 45 days
H100 from$2.60per GPU-hour, on-demand
Entry price$2.60cheapest GPU-hour listed
Countries11 named regions
Trust Score v0.171/100
- Company & track record47
- Infrastructure control100
- Pricing & billing transparency60
- Reliability signals70
- Security & compliance80
- Customer signal—
Scored Sep 29, 2026 from the provider's own pages. Dimensions not yet measured are excluded from the denominator. Criteria ↓
Trust Score · 5 of 6 dimensions measuredThe score, criterion by criterion
Each point comes from something the provider publishes, with the page it was read on and the date. Anything not disclosed earns zero.
Company & track record
weight 15%
47 / 100
| Founding year published | not disclosedgmicloud.ai · Sep 29, 2026 | 0 / 4 |
| Headquarters published | GMI Cloud US Inc.gmicloud.ai · Sep 29, 2026 | 3 / 3 |
| Legal entity named | GMI Cloud US Inc.gmicloud.ai · Sep 29, 2026 | 4 / 4 |
| Funding or ownership disclosed | not disclosedgmicloud.ai · Sep 29, 2026 | 0 / 4 |
Infrastructure control
weight 15%
100 / 100
| Capacity model published | operates or rents dedicated capacitygmicloud.ai · Sep 29, 2026 | 10 / 10 |
| Regions published | 1 regions or countriesgmicloud.ai · Sep 29, 2026 | 5 / 5 |
Pricing & billing transparency
weight 20%
60 / 100
| Public prices without an account | yes (read on the pricing page)gmicloud.ai · Sep 29, 2026 | 5 / 5 |
| Billing granularity published | per minutegmicloud.ai · Sep 29, 2026 | 4 / 4 |
| Egress policy published | not disclosedgmicloud.ai · Sep 29, 2026 | 0 / 4 |
| Stopped-instance billing published | not disclosedgmicloud.ai · Sep 29, 2026 | 0 / 4 |
| Public IPv4 policy published | extra: Elastic IP billed per minute (pay as you go), price shown only in the console; disassociated EIPs keep billing until releasedgmicloud.ai · Sep 29, 2026 | 3 / 3 |
Reliability signals
weight 20%
70 / 100
| Status page | not disclosedgmicloud.ai · Sep 29, 2026 | 0 / 6 |
| SLA published | Monthly uptime >= 99.9% for instances in multiple regions, >= 99% for instances in a single region; service credits in GPU-hours (GPUs impacted x downtime), capped at the monthly feegmicloud.ai · Sep 29, 2026 | 6 / 6 |
| SLA with a numeric uptime commitment | Monthly uptime >= 99.9% for instances in multiple regions, >= 99% for instances in a single region; service credits in GPU-hours (GPUs impacted x downtime), capped at the monthly feegmicloud.ai · Sep 29, 2026 | 3 / 3 |
| Uptime history published | About page states 'Platform Availability 99.99%'; Developers page states a '99.9% Uptime target'gmicloud.ai · Sep 29, 2026 | 5 / 5 |
| Observed availability (30 days) | not measured yet | — |
Security & compliance
weight 15%
80 / 100
| SOC 2 | Type IIgmicloud.ai · Sep 29, 2026 | 8 / 8 |
| ISO 27001 | yes (certificate listed in the trust center)gmicloud.ai · Sep 29, 2026 | 4 / 4 |
| HIPAA or GDPR DPA | not disclosedgmicloud.ai · Sep 29, 2026 | 0 / 3 |
Customer signal
weight 15%
not measured
| Aggregated public reviews | not collected yet | — |
provisional v0.1: evidence from the provider's own pages; dimensions not yet measured are excluded from the denominatorFull method →
Prices by GPU · 4 GPU types with a current offerWhat GMI Cloud charges, against the market
Cheapest current on-demand true cost here, next to the market median for the same GPU. Negative means cheaper than the market.
| GPU | Cheapest here | Market median | vs market | Current offers | Modes | |
|---|---|---|---|---|---|---|
| H100 | $2.60 | $3.17 | −18% | 1 | on-demand | H100 prices |
| H200 | $3.20 | $4.80 | −33% | 1 | on-demand | H200 prices |
| B200 | $5.00 | $7.50 | −33% | 1 | on-demand | B200 prices |
| GB200 | $8.00 | $8.00 | −0% | 1 | on-demand | GB200 prices |
All offers · 14 readings on fileEvery GMI Cloud offer, with its receipt
Current offers (last 45 days) ranked by true cost, all rental modes, each with its read date. Older readings stay listed below, in grey.
| GPU | Configuration | Mode | Region | Stock | List | True cost | Read | |
|---|---|---|---|---|---|---|---|---|
| H100SXM | 1× H100 SXM | On-demand | — | Stock unknown | $2.60 | $2.60 | 00:32 UTCfresh | Receipt |
| H200 | 1× H200 | On-demand | — | Stock unknown | $3.20 | $3.20 | 00:32 UTCfresh | Receipt |
| B200 | 1× B200 | On-demand | — | Stock unknown | $5.00 | $5.00 | 00:32 UTCfresh | Receipt |
| GB200 | 1× GB200 | On-demand | — | Stock unknown | $8.00 | $8.00 | 00:32 UTCfresh | Receipt |
4 fresh offers · prices per GPU-hour · true cost includes the add-ons the provider publishesOpen in the comparison tool →
True cost & billing · read Sep 29, 2026What the hourly price does not say
Billing granularity, minimums, what a stopped machine costs, traffic, addresses, deposits and identity checks, as published by the provider.
Billing
- Billing granularity
- per minute
- Minimum charge
- not published
- Stopped instance
- not published
- Egress
- not published
- Public IPv4
- extra: Elastic IP billed per minute (pay as you go), price shown only in the console; disassociated EIPs keep billing until released
- Billing model
- both
- Minimum deposit
- not published
- Free credits
- Coupons and a referral program that grants credits; no amount published
Account
- Payment
- card · invoice · crypto (not published)
- Sign-up
- email, password, export-control certification (ownership, end use, access locations on request)
- Identity check
- not published
- Support
- email, discord, console support contact
- Enterprise support
- not published
- SLA
- Monthly uptime >= 99.9% for instances in multiple regions, >= 99% for instances in a single region; service credits in GPU-hours (GPUs impacted x downtime), capped at the monthly fee
- Status page
- not published
- Published uptime
- About page states 'Platform Availability 99.99%'; Developers page states a '99.9% Uptime target'
Infrastructure
- Regions
1 named regions in 1 countries
us-denver-1 (United States)- Countries
- United States
- Platform
- API, serverless, Jupyter, SSH, custom Docker images, bare metalSDK: Python and TypeScript clients for the OpenAI-compatible inference API; GMI Sandbox Python SDKKubernetes: managedInterconnect: RDMA-ready networking (no bandwidth figure published)Clusters: Managed GPU Cluster Service (Kubernetes) with H200 or B200 nodes, requested in the console and activated by support; GB200/GB300 NVL72 via sales
- Storage
- not published
- Migration out
- Containers run from templates or custom images, so workloads are portable as Docker images. Reconfiguring a container wipes its data, so data must be copied out (SSH, Jupyter or web shell) before changes or termination. Elastic IPs are not released automatically on termination. Prepaid-plan containers cannot be terminated before their rental period ends.
Security & compliance
- SOC 2
- Type II
- ISO 27001
- yes (certificate listed in the trust center)
- HIPAA
- not published
- GDPR / DPA
- not published
- Other
- Sanctions and Export Control Compliance Policy; export control certification in the Console Terms of Service
- Legal entity
- GMI Cloud US Inc.
- Funding
- not published
Read on the provider's own pages · 24 pages read · facts marked "not published" were looked for and not foundPages read ↓
Assessment · documents read Sep 29, 2026GMI Cloud, in short
What stood out when we read the documents: strengths, weaknesses, who it suits, and what we could not find.
Strengths
- Published per-GPU-hour 'from' prices: H100 2.00, H200 2.60, B200 4.00, GB200 NVL72 8.00 USD
- GB200 NVL72 listed as available now and GB300 NVL72 on pre-order
- SOC 2 Type 2 report and ISO 27001 certificate listed in a public trust center
- Written SLA: 99.9% multi-region, 99% single-region, with GPU-hour service credits
- Serverless inference API (OpenAI-compatible, 200+ models) and dedicated endpoints on the same account
- Managed Kubernetes GPU clusters and bare metal with root access
Weaknesses
- Pricing page shows only 'from' prices; exact SKU rates, storage, Elastic IP and networking prices are visible only in the console
- Container and bare-metal access is gated per organization and must be enabled by support; cluster requests are not processed automatically
- No interruptible or spot tier published
- Prepaid credits, top-ups and coupons are non-refundable under the Console Terms of Service
- Only one data-center identifier (us-denver-1) is named in public docs despite 'US, APAC and EU' claims
- No public status page found
Ideal for
- Teams that want H200, B200 or GB200 capacity under a sales-negotiated commitment
- Inference-first teams combining a serverless model API with dedicated GPU capacity
- Buyers who need SOC 2 Type 2 and ISO 27001 evidence
Not published
- H100 form factor (SXM or PCIe) and VRAM for all GPUs: not stated on the pricing or GPUs pages
- Whether the 'from' prices are on-demand or committed rates: not stated
- Reserved or commitment prices and terms: sales only
- Storage, egress and Elastic IP prices: console only, not published
- Stopped-instance billing for bare metal: not documented
- Payment methods beyond saved cards (auto-pay) and invoices
- crypto acceptance: not published
- Identity verification at sign-up: only email verification is documented
- export-control information may be requested
- Full region list: only 'us-denver-1' is named in API docs
- marketing mentions US, APAC and EU without locations
- HQ city, founding year and funding: not stated on the pages read
- CLI and Terraform provider: not found in the docs index
Observations
Pricing page shows five GPU cards with 'from $X /GPU-hour' and no configuration details (vCPU, RAM, disk)
Pricing FAQ says networking and storage are provisioned separately and points to sales for detailed pricing
Docs pricing page states that rates live only in the console (Compute > Home)
Primary calls to action are 'Contact Sales' and 'Start in Console'
Pages read
- gmicloud.ai/en/pricing — GPU prices, pricing FAQ (per GPU-hour, dedicated, networking/storage separate, commitments
- gmicloud.ai/en/gpus — GPU catalogue, GB200/GB300 NVL72 labels, product types
- gmicloud.ai/en — H100/H200 prices, serverless inference
- gmicloud.ai/en/company/about — 30,000+ GPUs, 99.99% availability claim, SOC 2 / ISO 27001 claim, 24/7 support
- gmicloud.ai/en/developers — Inference API, SDKs, 99.9% uptime target
- docs.gmicloud.ai/cluster-engine/index — Product formats, request flow, product entitlements
- docs.gmicloud.ai/cluster-engine/resources/managed-gpu-clusters — Per-minute pay-as-you-go billing, prepaid option, manual activation
- docs.gmicloud.ai/cluster-engine/resources/containers — SSH/Jupyter access, prepaid renewal periods (1 day to 3 years), termination rules
- docs.gmicloud.ai/cluster-engine/resources/bare-metal — Bare metal management, gated access
- docs.gmicloud.ai/cluster-engine/resources/elastic-ip — IPv4 billed per minute, charges while disassociated
- docs.gmicloud.ai/cluster-engine/resources/templates — Built-in templates
- docs.gmicloud.ai/cluster-engine/user-management/account — Auto-pay with saved card, usage tiers
- docs.gmicloud.ai/cluster-engine/user-management/credits-coupons — Credit balance, coupons, referral
- docs.gmicloud.ai/cluster-engine/user-management/billing — Orders and receipts
- docs.gmicloud.ai/inference-engine/billing/price — Rates shown only in the console
- docs.gmicloud.ai/api-reference/introduction — REST API, email verification at sign-up, us-denver-1 IDC
- docs.gmicloud.ai/inference-engine/marketplace/dedicated — Billing only while Running
- gmicloud.ai/en/legal/service-level-agreement — SLA figures and service credits
- gmicloud.ai/en/legal/services-agreement — Monthly invoicing in advance in USD, 10-day payment terms
- gmicloud.ai/en/legal/console-terms-of-service — Legal entity, prepaid account credits, non-refundable credits, export control certification
- gmicloud.ai/en/legal/sanctions-and-export-control-compliance-policy — Sanctions representations
- gmicloud.ai/en/legal/privacy — Card data collected via payment processor, data stored in the US and Taiwan
- trust.gmicloud.ai/ — SOC 2 Type 2 report, ISO 27001 certificate
- gmicloud.ai/en/blog/gmi-cloud-commits-500-million-to-expand-ai-infrastructure-for-frontier-ai-customers — Headquarters
Questions about GMI Cloud
What GPUs does GMI Cloud offer?
GMI Cloud has 4 GPU types with a fresh offer today, for example H100 from $2.60, H200 from $3.20, B200 from $5.00, GB200 from $8.00 per GPU-hour on-demand. The full list with prices and market medians is on this page.
How does GMI Cloud bill?
Billing granularity: per minute.
How trustworthy is GMI Cloud?
Its Trust Score is 71/100 (v0.1), read from its own documents. Strong on infrastructure control. Weaker on company & track record, pricing & billing transparency. The score measures transparency and published guarantees, not customer satisfaction; every criterion and its source is listed above.
Where are GMI Cloud's data centres?
1 named regions in 1 countries: United States.
How do you pay GMI Cloud, and are there credits?
Payment by card, invoice, crypto. Free credits: Coupons and a referral program that grants credits; no amount published.