CoPilot Interview's model picker lists nine AI models: five free ones, with Mercury 2.5 · Inception as the default, two that need Pro (Gemini 2.5 Flash-Lite and GPT-6 Luna) and two that need Pro+ (Gemini 3 Flash and Claude Haiku 5.5). You choose one on the setup screen and can switch it mid-session from the 🤖 AI 1 menu. This page is the reference list; for which model suits which round, see which AI model to pick.
All nine models
| Label in the picker | Model family and provider | Plan needed |
|---|---|---|
| Mercury 2.5 · Inception ⚡⚡⚡ Instant | Mercury, by Inception (the default) | Free |
| GPT-OSS 120B · Cerebras ⚡⚡⚡ Instant | GPT-OSS, run on Cerebras | Free |
| Qwen3.8 27B · Cerebras ⚡⚡⚡ Vision | Qwen, run on Cerebras | Free |
| GPT-OSS 120B ⚡⚡ Fastest | GPT-OSS, run on Groq | Free |
| Qwen3.6 27B ⚡⚡ Fastest | Qwen, run on Groq | Free |
| Gemini 2.5 Flash-Lite ⚡ Fast | Gemini, by Google | Pro |
| GPT-6 Luna ⚡ Fast | GPT, by OpenAI | Pro |
| Gemini 3 Flash 🧠 Smart | Gemini, by Google | Pro+ |
| Claude Haiku 5.5 ⚡🧠 Smart | Claude, by Anthropic | Pro+ |
The ⚡ and 🧠 marks and the words Instant, Fastest, Fast, Vision and Smart are the app's own labels. Newer builds may show shorter family names, such as Mercury, GPT, Gemini and Claude.
Which plans include which models
| Plan | Models |
|---|---|
| Free ($0) | The five free models |
| Standard ($24.99 a month) | The five free models |
| Pro ($39.99 a month) | The free models plus Gemini 2.5 Flash-Lite and GPT-6 Luna |
| Pro+ ($49.99 a month) | All nine, adding Gemini 3 Flash and Claude Haiku 5.5 |
| Hour packs (from $9.99, one-time) | The free models and the two Pro models, not the Pro+ models |
Locked models show "🔒 Pro" or "🔒 Pro+" in the pickers. Choosing one shows, for example, "GPT-6 Luna needs the Pro plan ($39.99/mo) — GPT-6 Luna and Gemini 2.5 Flash-Lite. The model was left unchanged." with an Upgrade button. Full plan details are on the pricing page.
Where you choose a model
| Mode | Picker | Notes |
|---|---|---|
| 🎯 Candidate | 🤖 AI Model | Dual AI Mode adds a 2nd AI Model picker, preset to GPT-OSS 120B · Cerebras |
| 🎧 Mock Interview | 🤖 Interviewer AI | One model asks and grades every question; no Dual AI Mode |
| 🎤 Interviewer / HR | 🤖 AI Model | Dual AI Mode with its own 2nd AI picker |
The main pickers start on Mercury 2.5 · Inception, and the app doesn't remember your choice between launches, so check it before each session.
Switch models during a session
- Click 🤖 AI 1 in the session's top bar; it shows the current model's name. Pick another from the list; locked ones carry a 🔒.
- Click 1x Single to switch to 2x Dual. A 🤖 AI 2 menu appears and answers show in two columns, one per model. Press ⚡ AI Answer to re-answer the last question both ways.
- Check the Model figure in the stats row to see what is answering. The saved session file also records the model used.
Assistants with two AI models explains when side-by-side answers help.
Use your own API key (optional)
You don't need one: the built-in models need no setup. If you want to run on your own provider account, turn on 🔑 Use my own API key on the Candidate setup screen, choose a Provider (Google Gemini, OpenAI, Anthropic, Groq or Cerebras), optionally type a model name, and paste the key; the field notes it is "stored locally only". Starting with the switch on and no key shows "Enter your API key, or turn off "Use my own API key"." The app doesn't say whether your own key changes your plan's session limits.
What each model receives
The model you picked receives the question text plus the résumé and job description you entered, over an encrypted connection through CoPilot Interview's API proxy. Audio goes separately to speech-to-text (Groq Whisper, with OpenAI as a fallback), and on paid plans the text being read aloud goes to the voice provider. What the AI model provider sees goes into more detail.
CoPilot Interview is an AI interview assistant for Windows and Mac, and the cheapest way to compare models is a free AI mock interview: answer the same three questions with two models and keep the one whose answers you would actually say.
Frequently asked questions
What is the default AI model in CoPilot Interview?
Mercury 2.5 by Inception. It is preselected in the Candidate, Mock Interview and Interviewer / HR pickers on every plan, including the free plan.
Do I need my own API key for CoPilot Interview?
No. The built-in models work through CoPilot Interview's own connection with no setup. Your own key is optional, under 🔑 Use my own API key on the Candidate setup screen.
Why did my model go back to Mercury?
The app doesn't remember the model between launches. Each time it opens, the pickers start on Mercury 2.5 · Inception, so choose again before you start.
Which model should I use for coding or system design?
That depends on the round and your plan. The guide to which AI model to pick in CoPilot Interview gives round-by-round advice; this page only lists the models and plans.
Compare plans and models
See what each plan includes, from the free plan to Pro+, with session lengths and prices in US dollars.
See pricing