OpenAI

gpt-4o-mini is the default and covers almost every dictation use case — fast, cheap, and well-calibrated for short cleanup tasks. Open Settings → Cleanup to pick from the model list or enter a custom identifier.

Standard models (text and vision-capable are marked 👁):

• gpt-4o-mini — fastest (cleanup default) 👁
• gpt-4o — highest quality 👁
• gpt-4.1 — current 👁
• gpt-4.1-mini — current, cheaper 👁
• gpt-4.1-nano — text-only, smallest

Reasoning models — slower, useful for complex reference-aware turns:

• o1 — reasoning 👁
• o1-mini — reasoning, text-only
• o3 — reasoning 👁
• o3-mini — reasoning, text-only
• o4-mini — reasoning 👁

Reasoning models send a different request shape (no temperature, max_completion_tokens instead of max_tokens). Parleq detects reasoning models automatically from the model name — no manual family switch needed, unlike Azure.

Per-token^† at OpenAI's published rates. Cleanup prompts are short (a few hundred input tokens, a few hundred output tokens), so a typical day of personal dictation runs to single-digit cents on gpt-4o-mini. Settings → Usage tracks per-call token counts and rolling cost estimates.

† Third-party pricing, free tiers, and quotas are set by the providers — AWS, Google, Microsoft, Okta, and others change them without notice. Cost statements here were accurate when written; verify current rates on the provider's pricing page before you build.

Calls go to OpenAI's API under your account's consumer or pay-as-you-go terms. If your organization needs Microsoft's enterprise data-privacy commitments or Azure's regional deployment guarantees, use Azure OpenAI — the same models on Microsoft infrastructure.

Audio never reaches OpenAI. Only the text transcript output by Parleq's on-device speech model is sent.