Denvr AI Cloud (AI Compute & AI Inference)

Offering overview

Operating its own data centers in Canada and the United States since 2022, Denvr sells accelerated compute in three shapes: bare metal, virtual machines, and containers, with on-demand instances billed per minute and reserved contracts discounted up to 20%. The fleet spans NVIDIA H200, H100, A100 (including 20 GB MIG slices), A40, and GH200 alongside Intel Gaudi 2, with B200, GB200, Gaudi 3, and AMD MI300X listed for reservation. Clusters scale to 1,024 GPUs over 3,200 Gbps InfiniBand. The AI Inference product takes a different tack from token-metered APIs: each endpoint runs on single-tenant GPUs behind an OpenAI-compatible API, so customers pay for the hardware hour rather than usage. A catalog of open-weight models (Llama 3.3, DeepSeek R1, GPT-OSS, Qwen3-Coder-Next, Gemma 3) is pre-tuned for context length and batch size, and custom or fine-tuned models can be served through vLLM or Ollama. REST APIs, a CLI, Python and Terraform bindings, managed NVMe shared storage, and a startup credit program round out the platform. Published pricing starts at $0.58 per GPU-hour for an A100 MIG 20 GB slice, with H100 SXM at $2.45 per GPU-hour; H200 is reserved-only.

Technical specifications

Containers

Unknown

Kubernetes

Unknown

Root access

Unknown

API access

Unknown

Autoscaling

Unknown

Persistent processes

Unknown

AI capabilities

Model serving

Llama 3.3, DeepSeek R1, GPT-OSS, Qwen3-Coder-Next, Gemma 3

AI frameworks

Not stated

Inference endpoints

Unknown

Fine-tuning

Unknown

Vector database

Unknown

Model catalog

Unknown

Monitoring

Not stated

Evaluation

Not stated

Security & compliance

Geography

Provider headquarters are separate from offering infrastructure and availability.

Available plans

Pricing can change. Promotional prices are shown separately from standard prices.

Questions & answers

What compute deployment options does Denvr offer?

Denvr offers bare metal, virtual machines and containers, with on-demand instances and reserved contracts. It operates its own data centers in Canada and the United States.

How are Denvr AI Inference endpoints deployed and billed?

Each endpoint runs on single-tenant GPUs behind an OpenAI-compatible API. Billing is based on GPU-hours rather than token usage.

Can Denvr serve custom or fine-tuned models?

Yes. Custom or fine-tuned models can be served through vLLM or Ollama. Denvr also provides a catalog of open-weight models.

What automation interfaces does Denvr provide?

Denvr provides REST APIs, a CLI, and Python and Terraform bindings.