Copilot vs. raw API: GitHub explains what you're actually paying for

GitHub published a detailed explanation of the strategic difference between GitHub Copilot and direct API access to large language models, aimed at developers asking why they should pay for Copilot when the underlying models are available through raw APIs.

The core distinction, according to GitHub, is about what work you need to own. Copilot bundles model access with deep integration into the entire development workflow: the editor, repository, pull requests, issues, terminal, and organization-level policies. A developer starting from a GitHub Issue, inspecting the repository, modifying files, running the test suite, and opening a pull request performs a workflow where the model call is just one step. Copilot connects all these surfaces, handling context selection, tool use, retries, and the path to a reviewed pull request. In contrast, raw API access is appropriate for teams building custom systems—product features, internal agent platforms, evaluation harnesses, or automation pipelines—where the team controls prompts, retrieval, routing, logs, security models, and billing.

GitHub's billing structure now makes this layering clearer. Copilot plans include a monthly allocation of GitHub AI Credits; metered usage is calculated from input, output, and cached tokens at the listed rate for the selected model. Importantly, code completions and Next Edit Suggestions remain included in paid plans, while AI Credits apply to more resource-intensive chat and agentic work. Cost per task depends on more than the listed token rate: context selection, tool use, retries, and the overall workflow path all affect token spend and whether work finishes efficiently.

For organizations with existing model provider contracts, Copilot offers Bring Your Own Key (BYOK), currently in public preview. BYOK lets developers make supported provider models available in Copilot Chat, Copilot CLI, and VS Code. Supported providers include Anthropic, AWS Bedrock, Google AI Studio, Microsoft Foundry, OpenAI, OpenAI-compatible providers, and xAI. Through BYOK, your provider takes over the token bill while GitHub continues to develop and maintain the tooling and integrations. This approach allows teams to keep an existing commercial relationship while developers use Copilot in their normal workflow. Copilot CLI also supports local and external BYOK models, including Azure OpenAI and local Ollama models. Enterprise and organization admins choose which of Copilot's supported 20+ models are enabled for their teams, whether GitHub-hosted or connected through BYOK.

GitHub's position is that the choice between Copilot and raw API access follows the work you need to own. Choose raw API access when building a system that requires custom behavior, integrations, and controls. Choose Copilot when the work is software development inside the tools and repositories where a team already writes, reviews, secures, and ships code. GitHub noted that shipping software involves work around the code—issues, pull requests, reviews, checks, actions, and security—and Copilot helps teams move through that work faster.

GitHub's post addresses a persistent developer question: why pay for Copilot when the same underlying models are available through direct API calls? The answer hinges on the distinction between model access and development workflow integration. Copilot operates as a layer above raw API access, bundling model calls with the infrastructure developers use daily—GitHub's editor integrations, repository context, pull request and issue management, terminal access, and organization-level controls. The recent billing change, which separated code completions (still included) from more resource-intensive chat and agentic work (now billed via AI Credits), makes this layering explicit.

For teams building internal products or custom automation, raw API access remains the correct choice because it grants full control over system design decisions: which files to retrieve, how to handle retries, where to store audit records, and which credentials to permit. These decisions require ownership by the development team. However, for teams focused on shipping software through their existing GitHub workflow, Copilot's integration advantage compounds—it reduces the tokens spent by providing rich repository and issue context automatically, eliminates the need to wire up organization policies separately, and makes usage visible through centralized billing dashboards rather than scattered API keys.

The Bring Your Own Key feature (in public preview) adds a third option: teams can maintain existing commercial relationships with model providers while accessing Copilot's integrated workflow. This flexibility suggests GitHub is positioning Copilot not as a replacement for API access but as a workflow layer for development teams, leaving raw APIs for builders of custom systems.

No comments yet. Be the first to share your thoughts!