Gcore review

Facts from Gcore’s own pricing and docs pages, September 2026 — 53 read from a quoted sentence on the page we archived.

Summary

Gcore rents GPUs from ≈$1.23 per GPU-hour on demand, bills hourly (GPU), and asks no minimum commitment. Egress is free. Gcore does not publish an uptime SLA.

GDPR · DPAFree egress

From

≈$1.23/GPU-hr

Billing

Hourly (GPU); per minute (inference)

Minimum commitment

None on pay-as-you-go

Egress

Free

Uptime SLA

not stated

Registered in

🇱🇺

Visit Gcore ↗

On-demand pricing

NVIDIA H100 SXM5 80GB

×8, on demand · H100 SXM

≈$2.59/hr

Deploy

NVIDIA H200 SXM5 141GB

×8, on demand · H200 SXM

≈$3.12/hr

Deploy

NVIDIA L40S 48GBLowest

×1, on demand · L40S

≈$1.23/hr

Deploy

NVIDIA B300 SXM6 288GB

×8, on demand · B300 SXM

≈$6.85/hr

Deploy

NVIDIA A100 SXM4 80GB

×8, on demand · A100 SXM

≈$1.45/hr

Deploy

See all 9 Gcore offers →

What’s good

Lowest published on-demand rate is ≈$1.23/GPU-hr (NVIDIA L40S 48GB), with no minimum commitment

Egress is free — no per-GB charge on data leaving

Billing is hourly (GPU)

Self-serve signup — no sales call to get started

Free credits available: free trial periods offered

Commitment discounts: reserved and subscription pricing

Plans

	Bare metal GPU clusters	Virtual GPU	Everywhere Inference
From	€1.09/GPU-hr (L40S) Gcore · Bare metal GPU clusters Every on-demand rate we hold for Gcore, per GPU per hour, cheapest first. NVIDIA L40S 48GB ×1, on-demand · L40S ≈$1.23/hr Deploy NVIDIA A100 SXM4 80GB ×8, on-demand · A100 SXM ≈$1.45/hr Deploy NVIDIA H100 SXM5 80GB ×8, on-demand · H100 SXM ≈$2.59/hr Deploy NVIDIA H200 SXM5 141GB ×8, on-demand · H200 SXM ≈$3.12/hr Deploy NVIDIA B300 SXM6 288GB ×8, on-demand · B300 SXM ≈$6.85/hr Deploy Prices as last read from Gcore’s own pages. Open a card for every provider renting it.	€2.36/hr (1× H100, 24 vCPU / 232 GB / 1.25 TB)	€1.08/hr (1× L40S) Gcore · Everywhere Inference Every on-demand rate we hold for Gcore, per GPU per hour, cheapest first. NVIDIA L40S 48GB ×1, on-demand · L40S ≈$1.23/hr Deploy NVIDIA A100 SXM4 80GB ×8, on-demand · A100 SXM ≈$1.45/hr Deploy NVIDIA H100 SXM5 80GB ×8, on-demand · H100 SXM ≈$2.59/hr Deploy NVIDIA H200 SXM5 141GB ×8, on-demand · H200 SXM ≈$3.12/hr Deploy NVIDIA B300 SXM6 288GB ×8, on-demand · B300 SXM ≈$6.85/hr Deploy Prices as last read from Gcore’s own pages. Open a card for every provider renting it.
Minimum term	Per Service Order; pay-as-you-go available	None	None
Billing	Hourly rate per node; reserved terms	Hourly while powered on; shelving free of compute	Per minute of compute
Tenancy	Dedicated 8-GPU nodes; InfiniBand fabric	1–8× H100 VMs	Managed inference endpoints, autoscaling
Provisioning	—	Multi-node clusters in one API call	Fast cold starts
Payment	Postpaid invoice by default	Postpaid invoice by default	Postpaid invoice by default

Details

Billing & access

How you pay

Postpaid invoice by default; prepaid credits for some products; subscriptions

Currency

EUR, ex VAT

Commitment discounts

Reserved and subscription pricing

Free credits

Free trial periods offered; startup program

Refunds

Prepaid credits non-refundable; unused prepaid fees credited on final invoice

Late payment

1.75%/month plus statutory interest

Provisioning guarantee

Pay-as-you-go capacity is as-available

Under the pay-as-you-go tariff Gcore agrees not to reclaim services actively in use, but capacity is provided on an as-available basis.

What you get

GPUs listed

L40S, A100, H100, H200, B300, GB300

Regions

50+ cloud locations; 33 named on the cloud price list

Almaty, Amsterdam, Baku, Chester, Chicago, Darmstadt, Dubai, Frankfurt, Helsinki, Hong Kong, Incheon, Istanbul, Johannesburg, Lefdal, London, Luxembourg, Manassas, Mumbai, Paris, Santa Clara, São Paulo, Sines, Singapore, Sydney, Tashkent, Tokyo, Warsaw — GPU availability varies by location.

Node configuration

8× H100 SXM: 640 GB VRAM · 2 TB RAM · 224 vCPU · 8× 3.84 TB NVMe

Interconnect

InfiniBand 3,200 Gbps (H100/H200), 800 (A100), 3,600 (B300); NVLink on GB300

Local storage

NVMe per node (e.g. 8× 3.84 TB on 8× H100)

Persistent storage

Object storage (S3), file shares, block storage

Public endpoints

Protected inference endpoints; load balancers

Private networking

Private networks; L2 between bare metal and VMs

Images

Ubuntu Server 22.04 LTS

Access methods

Dashboard, API, Terraform

Model deployment

Everywhere Inference; custom software via APT

Managed inference

Everywhere Inference, per minute

Managed Kubernetes

Managed Kubernetes with GPU nodes; control plane free

Serverless

Function as a Service; Everywhere Inference

Kubernetes integration

Managed Kubernetes with GPU node pools

Terraform

Yes

FAQ

How much does Gcore cost?

As of September 2026, the lowest published on-demand rate is ≈$1.23 per GPU-hour for the NVIDIA L40S 48GB. Gcore publishes an on-demand rate for 5 GPU models. Billing is hourly (GPU).

Does Gcore have an uptime SLA?

Gcore does not publish an uptime commitment on its own site.

Is there a minimum commitment on Gcore?

No. Gcore states none on pay-as-you-go; subscription terms billed in full.

Does Gcore charge for egress?

No. Gcore states that egress is free.

Can I sign up to Gcore without approval?

Gcore states that signup is self-serve.

Are Gcore payments refundable?

Gcore states: prepaid credits non-refundable; unused prepaid fees credited on final invoice.

Where are Gcore's data centres?

Gcore lists 50+ cloud locations; 33 named on the cloud price list.

Who do you contract with at Gcore?

The contracting entity is G-Core Labs S.A. (Luxembourg).

Reviewed September 2026, from Gcore’s own site. 53 facts, each quoted from the page it was read on.