@tifan/pi-fast-mode
v0.2.0
Published
Toggle OpenAI Fast Mode per model and track response TPS.
Maintainers
Readme
@tifan/pi-fast-mode
Toggle OpenAI Fast Mode per model and track response TPS.
Install
pi install npm:@tifan/pi-fast-modeUsage
Run /fast to open the list of supported models. Select a model to toggle Fast Mode. Enabled models are marked ✓. The setting is saved per exact provider/model pair.
Run /tps to toggle response TPS. The setting is saved in the same configuration file.
When TPS is enabled, the status shows the latest response rate, median response rate, and median time to first token:
last 58 t/s · med 44 t/s | 2.1s ttftResponse TPS uses Pi's provider-reported output tokens divided by the time from turn start to assistant message end. It includes reasoning tokens and response wait time. It does not include time spent executing tools.
Fast Mode adds service_tier: "priority" for these exact models:
openai/gpt-5.4 openai-codex/gpt-5.4
openai/gpt-5.5 openai-codex/gpt-5.5
openai/gpt-5.6 openai-codex/gpt-5.6
openai/gpt-5.6-sol openai-codex/gpt-5.6-sol
openai/gpt-5.6-terra openai-codex/gpt-5.6-terra
openai/gpt-5.6-luna openai-codex/gpt-5.6-lunaConfiguration
Preferences are stored at $PI_CODING_AGENT_DIR/extensions/pi-fast-mode.json:
{
"models": ["openai-codex/gpt-5.6-luna"],
"tpsEnabled": true
}tpsEnabled defaults to true when it is missing. Unsupported models cannot be enabled.
Release notes
See CHANGELOG.md
