Yes. IBM Bob’s self-hosted deployment is generally available as of September 2026. Its backend runs on a customer-managed Red Hat OpenShift cluster, either on premises or in the customer’s own cloud account. The key qualification is that “self-hosted” does not automatically mean all AI inference stays inside your network: that depends on whether you run a supported model locally or send model requests and their code context to an approved external service.
What IBM means by “self-hosted”
IBM Bob’s self-hosted version puts the Bob backend on infrastructure your organization operates, rather than on IBM-managed SaaS infrastructure. IBM announced general availability on September 30, 2026; its October 1 product release says general availability took effect September 24. The deployment is supported on customer-managed Red Hat OpenShift Container Platform, on premises or in your own cloud account. IBM’s announcement and the October release post describe the launch.
Bob’s self-hosted offering includes its core IDE experience, BobShell, parallel tool calling, agent harness, skills, and modes in supported configurations. It is deployed with the Bob Kubernetes Operator and Helm charts, using separate operator and instance namespaces. IBM says a dedicated cluster is not required: Bob can share an appropriately resourced OpenShift cluster with other workloads. IBM’s self-hosted overview describes the platform and operating model.
Where inference happens—and whether code context leaves
The important privacy and sovereignty decision is the model connection, not just the location of Bob’s backend. In the hybrid route, Bob’s backend, identity, audit logs, and usage metering stay on your cluster, while model requests—including the code context in those requests—go to the external model service your organization has approved. That service’s terms and controls govern the external processing.
Free tools Windows power users keep installed
One-click scans. No signup required.
For the fully self-hosted route, the organization runs inference on its own GPU infrastructure. IBM’s October release post names NVIDIA Nemotron 3 Ultra and Poolside Laguna S 2.1 for this configuration. It also describes a small guardrail model that screens inputs and outputs. IBM says required capacity depends on factors including quantization, context length, and the number of developers using the system concurrently; the public announcement does not establish one universal GPU configuration.
IBM’s announcement also names NVIDIA Nemotron and Poolside Laguna as supported local-model families. Model availability and configuration should be confirmed against the deployment documentation and IBM account team for the intended setup. The release post provides IBM’s named models and deployment details.
Self-hosted versus IBM-managed SaaS
| Decision area | IBM-managed SaaS | Self-hosted |
|---|---|---|
| Infrastructure and upgrades | IBM manages the infrastructure and upgrades. | Your organization operates the OpenShift environment and the Bob deployment lifecycle. |
| Model and request context | Depends on the SaaS model configuration; the cited self-hosted documentation does not establish SaaS-specific data-flow details. | With a local model, inference runs on your infrastructure. With a hybrid external model service, requests and code context are sent to that service. |
| Disconnected use | The cited materials do not describe SaaS as an air-gapped option. | IBM supports disconnected installation, including image mirroring into a private registry. |
| Identity and platform security | IBM manages the SaaS infrastructure. | Your team configures networking, storage, identity, and platform security monitoring. IBM says security event logging and monitoring are handled at OpenShift platform level, not provided by Bob. |
IBM’s April 28, 2026 general-availability release described Bob as SaaS and said on-premises deployment was planned for the future. That was the position at the time; the September and October announcements supersede it for current availability. The April release is useful historical context.
Can IBM Bob run in an air-gapped environment?
IBM says the self-hosted deployment supports disconnected installation. The release post describes using bobctl to mirror images into a private registry; an organization can transfer the mirrored images from a connected system into the disconnected environment. This addresses installation without an ongoing connection to a public image registry; it does not by itself establish that every model, dependency, or operational workflow is available offline. Confirm the complete model and platform requirements for the intended air-gapped deployment with IBM.
Rank #3
IBM says a single-node OpenShift installation can support a proof of concept. That is not a production sizing recommendation: production capacity depends on workload and model choices, and IBM directs customers to its installation documentation for prerequisites and sizing.
What your team must operate
Self-hosting gives the organization control over the deployment environment, but also transfers platform work that IBM would manage in SaaS. The customer configures and operates networking, storage, identity, upgrades, and the underlying OpenShift environment. Security event logging and monitoring are platform-level responsibilities; Bob does not provide those capabilities itself.
For corporate identity, IBM’s documentation describes configuration through LDAP or Active Directory. Smaller installations can manage users directly through the included identity service. Choose the identity approach as part of deployment planning, alongside the organization’s own audit and monitoring controls.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Access, licensing, and what is not publicly specified
IBM describes access as sales-led: contact IBM or an IBM Business Partner, or use the Contact Sales route. The account team arranges the organization’s entitlement. Pricing and exact enterprise entitlement terms are not stated in the reviewed public materials, so a prospective customer will need to confirm them directly.
IBM identifies optional Premium Packages for Java Modernization, IBM i, and IBM Z. Their availability is subject to applicable licensing and deployment conditions. Neither the public release materials nor the overview establish production minimums or one universal sizing prescription.
Why the deployment option matters
IBM frames the launch around organizations that need more control over where AI operates and how sensitive code is handled. In its October 1, 2026 press release, IBM attributed a finding that 68% of surveyed executives say meeting data residency and sovereignty requirements across geographies is challenging to the IBM Institute for Business Value report The Calculus of AI Sovereignty, published in June 2026. This is a survey finding, not evidence that self-hosting alone satisfies any particular organization’s legal or regulatory obligations. IBM’s October 1 release also quotes its GM of AI and Automation, Neel Sundaresan: “Organizations need AI that operates inside environments they have control over, especially when working with sensitive code and regulated data.”
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.
