GitHub Copilot Ships Review Effort Levels, Adds Kimi K3

GitHub Copilot moved review effort levels to general availability on Aug. 7 and added Kimi K3 as a selectable model on Aug. 6, per the changelog.

Effort levels turn review depth into a per-invocation dial instead of a flat cost. It's the same knob Claude Code and Codex added earlier this year. Engineering teams choosing a review harness are comparing a control they already expect rather than a differentiator.

Three coding harnesses expose that same control, which makes it closer to baseline than to a selling point. The changelog also mentions an ROI section landing in the impact dashboard, giving teams a place to see what the effort spend is buying. That's the pairing that makes sense. Once you can dial cost up or down, you need a way to see if it paid off.

Kimi K3 is the more interesting line item. Copilot picking it as a selectable option says GitHub is competing on model breadth as much as on review features.

With effort levels standard across Claude Code, Codex, and Copilot, the three harnesses aren't differentiating on depth control anymore. They're differentiating on which models you're allowed to point that control at.

Watch whether Copilot keeps adding non-default models faster than it adds new review controls. That's the tell for where GitHub thinks the competition actually is.