Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesiTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
NVIDIA has released the complete Run:ai platform as open source. Announced on October 13, 2026 as part of the NVIDIA DSX AI Factory Platform, the release lets teams inspect, build, and adopt the software without buying a Run:ai subscription. NVIDIA also distributes an integrated Kubernetes stack through NVIDIA NGC as a Community Product. That package requires acceptance of NVIDIA’s applicable license terms, but not a paid subscription or NVIDIA support entitlement. Enterprise support, managed operations, and other separately contracted services remain outside the free offering.
What NVIDIA released
The release covers four independently adoptable components, so an organization can deploy only the capabilities it needs or use the integrated stack:
| Component | Primary function |
|---|---|
| KAI Scheduler | GPU-aware scheduling for Kubernetes workloads. |
| GPU Fractioning Operator | CUDA-native sharing of GPUs between workloads. |
| KAI Resource Management | Organizational quotas, hierarchy, and resource governance. |
| Workload Map | A consistent contract for managing workloads across the platform. |
The source is available in NVIDIA’s Run:ai repository within the DSX AI Factory GitHub organization. NVIDIA says teams may inspect the code, contribute changes, build their own distributions, or adopt components independently. KAI Scheduler is explicitly licensed under Apache 2.0 and has been contributed to the Cloud Native Computing Foundation as a Sandbox project. Other components use the applicable license for each repository.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteNVIDIA also identifies Model Streamer and Grove as complementary open projects: Model Streamer addresses model loading, while Grove supports orchestration of multi-component inference workloads. They are related projects, not additional Run:ai subscription tiers.
#1 Best Overall
- NVIDIA Volta GV100 Architecture — 4,608 CUDA Cores, 640 1st-Gen Tensor Cores delivering 14 TFLOPS FP32 and 112 TFLOPS deep learning performance for AI training, inference, HPC, and scientific computing workloads
- 32GB HBM2 ECC Memory — 900 GB/s Bandwidth — High-bandwidth memory on a 4096-bit bus with ECC error correction provides the memory capacity and throughput required for the largest AI models, simulations, and datasets
- PCIe 3.0 x16 Interface — 250W TDP — Standard PCIe Gen3 connectivity with passive cooling designed for enterprise rack server deployment in HPE ProLiant, Dell PowerEdge, and Supermicro platforms with adequate chassis airflow
- NVLink — Scale to 96GB Unified Memory — Connect two V100 GPUs via NVLink at 300 GB/s bi-directional bandwidth to scale GPU memory from 32GB to 96GB for larger AI training and HPC workloads
- Multi-Precision Computing — Supports FP64 (7 TFLOPS), FP32 (14 TFLOPS), FP16 (112 TFLOPS) and INT8 precision modes for flexible deployment across training, inference, and scientific simulation workloads
Is Run:ai free now?
For the open-source repositories, there is no Run:ai subscription charge to inspect, modify, build, or deploy the code. The NGC Community Product likewise does not require a paid Run:ai subscription or a paid NVIDIA support entitlement.
NVIDIA’s announcement states: “No paid NVIDIA Run:ai subscription or NVIDIA support entitlement is required for the Community Product, and NVIDIA enterprise support is not included.” That distinction is the practical boundary: the software package is available without the former commercial entitlement, while formal enterprise assistance is not bundled.
What the NGC Community Product includes
NGC packages the integrated platform for deployment, including a Helm chart. This is useful when a team wants the four components together rather than assembling repositories and release artifacts independently.
- Choose the distribution model. Use the repositories when you need source-level control or only selected components; choose the NGC Community Product when you want the integrated stack.
- Review the applicable terms. NGC users must accept the NVIDIA Software License Agreement and Product-Specific Terms for the relevant NVIDIA AI products.
- Prepare the platform. Provide a compatible Kubernetes environment and NVIDIA GPU infrastructure, then assign responsibility for cluster security, upgrades, GPU drivers, and capacity planning.
- Deploy and configure. Install the NGC Helm chart for the integrated package, or deploy the chosen component repositories using their documented Kubernetes procedures.
- Set governance policies. Configure scheduling behavior, GPU sharing, quotas, organizational hierarchy, and workload-management rules for your users and teams.
- Operate it as your own platform. Monitor failed workloads, GPU utilization, upgrades, and compatibility; community or internal expertise replaces an included enterprise support contract.
Accepting NGC terms is a legal licensing step, not proof that a fee has been charged. The source repositories remain governed by the open-source license applicable to each component.
What still costs money
“Open source” does not make every service around the platform free. NVIDIA separates open-source software and free Community Products from paid Enterprise Products. Its general enterprise software model can include several structures:
| Commercial model | How it is typically charged | Run:ai-specific figure published? |
|---|---|---|
| Annual subscription | Per seat or per GPU-year. | No price listed in the October 2026 Run:ai announcement. |
| Consumption or pay-as-you-go | Often per GPU hour through a cloud marketplace. | No price listed. |
| Concurrent-user licensing | Based on the number of simultaneous users for products that offer this model. | No price listed. |
The available announcement does not publish a Run:ai subscription price, paid feature tier, or enterprise-support rate. Do not treat a generic NVIDIA licensing model as a Run:ai quotation. Any paid entitlement, support agreement, cloud infrastructure bill, or managed-service contract must be confirmed with the relevant vendor.
Rank #3
- Professional GPU with Blackwell Architecture in Compact Small Form Factor (SFF)
- Blackwell Architecture
- 24GB GDDR7 with PCIe 5.0 & Ray Tracing
- AI Workstation
Self-managed source versus packaged deployment
| Decision | Self-managed repositories | NGC Community Product |
|---|---|---|
| Control | Maximum control over code, components, and build process. | Integrated NVIDIA package with Helm deployment. |
| Entry cost | No Run:ai subscription charge stated. | No paid Run:ai subscription or NVIDIA support entitlement required. |
| Operational work | You select versions, assemble components, and maintain deployments. | You still operate Kubernetes and GPUs, but the stack is packaged together. |
| Support | Community, internal, or separately purchased assistance. | Community access; enterprise support is not included. |
| Legal step | Follow each repository’s applicable open-source license. | Accept the applicable NVIDIA Software License Agreement and Product-Specific Terms. |
Where partners and cloud charges fit
NVIDIA names eight DSX ecosystem partners integrating the open platform: Crusoe, Mirantis, Nebius, OpenNebula Systems, Rafay, Saturn Cloud, Spectro Cloud, and vCluster. Their offerings may provide managed GPU-cloud or AI-factory operations, but the announcement does not state partner prices, service-level agreements, reseller terms, or commissions.
A partner’s managed service can therefore add charges even when the underlying Community Product does not require a Run:ai subscription. The same applies to GPU instances, storage, networking, support labor, and other cloud consumption. Verify those costs and responsibilities directly with the provider.
Who should use which option?
- Use the repositories if your platform team needs component-level adoption, code changes, or a custom build and can own Kubernetes operations.
- Use the NGC Community Product if you want NVIDIA’s integrated stack and Helm-based packaging without purchasing a Run:ai subscription.
- Choose a managed partner service if your priority is outsourced GPU-cluster or AI-factory operations, and obtain a written quote for infrastructure, management, and support.
- Budget separately for enterprise support when response-time commitments, vendor escalation, or contractual assistance are mandatory.
The practical answer for buyers
NVIDIA Run:ai is now open source in the meaningful software sense: the complete platform’s source is available, its modules can be adopted separately, and the NGC Community Product is available without a paid Run:ai subscription or NVIDIA support entitlement. The free boundary ends at included enterprise assistance and other commercial services. Because NVIDIA has not published a Run:ai-specific price list with this launch, any dollar amount presented as the product’s official price would be unsupported.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
