Codex CLI: Persistent Fast Mode Toggle & Service Tier Selection

Codex CLI Gains Persistent Fast Mode and Flex Service Tier

Codex CLI 0.110.0 introduces a persistent /fast toggle — a new TUI command that switches the active service tier between fast and flex modes and retains the preference across sessions.

Fast vs. Flex Service Tiers

The two tiers represent different operating points on the inference speed-and-cost spectrum:

• Fast mode prioritizes low latency, suited for interactive development sessions where quick agent response times reduce friction in the coding loop.
• Flex mode uses a more cost-efficient execution path, suited for longer background tasks or batch agent work where latency is a secondary concern relative to cost.

The app server in 0.110.0 adds explicit support for both tiers, meaning the selected preference can propagate through connected integrations and enterprise app-server deployments — rather than being honored only by the local CLI process.

Usage

The toggle is invoked via the /fast slash command in the TUI. Activating it switches the session to fast mode; running it again returns to flex. Unlike earlier session-scoped settings in the CLI, the toggle state is persisted in the CLI's configuration layer and survives session restarts without requiring re-entry.

This is the first time Codex CLI has surfaced an explicit service tier control in the terminal interface. Previously, tier selection was opaque to end users, handled implicitly by the backend. The /fast toggle gives developers a direct, visible mechanism to manage the speed-cost tradeoff based on the nature of the work at hand.