GPT Fast Pi

This is the MVP Lab Pi extension for GPT Fast mode. It replaces the third-party @tunnckocore/pi-gpt-fast-mode package.

Behavior

When enabled, requests for these model IDs receive:

{
  "service_tier": "priority"
}

Matching is provider-agnostic. For example, all of these qualify:

mvp-lab/gpt-6-astra
openai/gpt-6-astra
openai-codex/gpt-6-astra
some-proxy/gpt-6-astra

Supported suffixes:

• gpt-6-astra
• gpt-5.4
• gpt-5.4-mini
• gpt-5.5
• gpt-5.6
• gpt-5.6-sol
• gpt-5.6-terra
• gpt-5.6-luna

The selected Pi model must also match the model field in the serialized provider payload. Other models are left unchanged.

Use

/fast

Run /fast again to turn it off. The default state is controlled by the global Pi settings:

{
  "pi-gpt-fast-mode": {
    "enabled": true
  }
}

The default shortcut is ctrl+alt+m. The extension is intentionally local and has no runtime dependency on a third-party Fast mode package.

Session persistence

Each toggle appends a non-LLM custom session entry:

gpt-fast-pi.state { version: 1, enabled: true|false }

When the session is reloaded, resumed, or forked, the latest state from the active branch is restored. If the session has no state entry, the extension falls back to the global pi-gpt-fast-mode.enabled setting.