Back to rankings

GitHub Copilot: Ollama as a BYOK Provider in JetBrains

GitHub CopilotView original changelog

GitHub Copilot for JetBrains IDEs now supports Ollama as a bring-your-own-key (BYOK) model provider, letting developers route chat and agent requests to locally hosted, open-weight models instead of GitHub's hosted lineup. The integration allows model selection and provider configuration throughout the JetBrains experience. It extends GitHub's 2026 BYOK expansion — which already covers custom OpenAI-compatible endpoints, the Copilot CLI, and the standalone Copilot app — with the first fully local, no-API-key-required option for JetBrains.

Key Takeaways

  • Ollama is now a selectable BYOK provider in GitHub Copilot for JetBrains, alongside the existing custom-endpoint BYOK option.
  • Because Ollama runs entirely on local hardware, code and prompts sent through it never leave the developer's machine.
  • The feature removes the per-token cost of hosted models for developers willing to run open-weight models locally.
  • It's the first no-API-key BYOK option for JetBrains — prior BYOK additions (April, July) still required a configured API key or custom endpoint.
  • Model selection and provider setup happen inside the same JetBrains UI Copilot already uses, with no separate plugin required.
  • The rollout follows a consistent 2026 pattern of GitHub extending BYOK support surface-by-surface — CLI in April, VS Code Business/Enterprise in April, the standalone app in June, and now JetBrains's local-model option in August.

Ollama Joins JetBrains as a BYOK Provider

GitHub Copilot for JetBrains IDEs now supports Ollama as a bring-your-own-key (BYOK) model provider, giving developers a way to route Copilot chat and agent requests to locally hosted, open-weight models instead of — or alongside — GitHub's hosted model lineup.

Why It Matters

Ollama runs models directly on a developer's own machine, so teams that need to keep code and prompts off third-party infrastructure, or that simply want to experiment with open-weight models at no per-token cost, can now do so from inside the same JetBrains chat and agent interface they already use for Copilot. The integration lets developers select Ollama-hosted models and configure the provider connection throughout the JetBrains experience, the same way they would configure any other BYOK provider.

Part of a Broader BYOK Push

The addition builds on GitHub Copilot's steady expansion of BYOK support across surfaces in 2026, including custom OpenAI-compatible endpoints for JetBrains introduced in July, BYOK for the Copilot CLI, and BYOK support in the standalone GitHub Copilot app. Ollama is the first fully local, no-API-key-required option added to that lineup for JetBrains, making it a natural fit for developers prioritizing data locality or cost control over frontier-model performance.