October 28 catch-up
GitHub announced image creation and editing inside Copilot in VS Code, its app, and CLI. The reasoning model calls a dedicated GPT Image 2.5 Flare tool; a preview appears and the asset is saved to the project. This is public preview. [1]
Access and cost are separate checks
The announcement names Pro, Pro+, Business, and Enterprise. Current documentation also lists Max and excludes Free and Student. Organization or enterprise model policy can disable generation. Check your plan and administrator’s policy. [1][2]
Text and image tokens for this tool consume AI credits separately from the conversation model. The pricing table uses token rates, not a flat price for each finished picture. A predictable per-image bill cannot be inferred without usage details. [2][3]
A visible label is optional
The visible Made with AI watermark is optional, initially off, and not centrally enforceable. Screenshots or re-encoding can remove C2PA metadata. We did not test an export. [2]
Our take
For experienced technicians making interface assets, the convenience is real in principle: fewer steps between a brief and a project file. Keep the asset’s budget owner and provenance decision explicit. Check lettering, logos, and the intended crop before using a result; an attractive preview alone is not a quality review.
Use an approved, non-sensitive practice brief first. Decide how the team will label and retain source files before the asset enters a customer-facing project.