On Peec AI's self-serve plans, you track three AI engines chosen from a fixed set of six. Perplexity costs extra. DeepSeek, Qwen and Claude appear only on Enterprise and on the agency Scale plan and above, where Peec's pricing tables mark them "(API)".
Which engines can you pick on Starter, Pro and Advanced?
Each self-serve plan lets you track three engines at a time. You choose them from the same six on every plan: ChatGPT, Google AI Mode, Google AI Overviews, Microsoft Copilot, Gemini and Naver AI (Peec pricing).
Peec calls these engines "models". An "active model" is one you have switched on for a project. Peec's docs say three are selected by default, and you can change the selection per project (Peec docs).
Plan limits count answers, not engines. Peec calculates monthly AI answers as tracked prompts × active models × tracking frequency; its pricing FAQ example uses daily runs over 30 days (25 × 3 × 30 = 2,250 answers). Active models are fixed at three on self-serve plans, so a fourth engine, bought as an add-on, adds both answers and cost.
Is Perplexity a paid add-on?
Yes, on the self-serve plans. Perplexity is not in their engine lists; it is offered as an "Additional Model" add-on, priced per extra engine.
The add-on price rises with the plan, because it reflects the number of prompts the plan includes (Peec pricing).
| Plan | Base price (monthly) | Engines included | One extra engine (monthly billing) | One extra engine (annual billing) |
|---|---|---|---|---|
| Starter | $95 | 3 of 6 | $35 | $30 |
| Pro | $245 | 3 of 6 | $85 | $70 |
| Advanced | $495 | 3 of 6 | $165 | $140 |
| Enterprise | Custom | All listed | Included | Included |
Source: peec.ai/pricing, checked 31 October 2026.
The add-on list contains ChatGPT, AI Mode, AI Overviews, Copilot, Perplexity and Gemini. So an add-on can be Perplexity, or a fourth engine from the base set. Naver AI is not in the add-on list. Peec does not say whether an add-on applies to one project or to all of them.
On Enterprise and on the agency Scale and Comprehensive plans, Perplexity is part of the included engine list (agency pricing).
Does Peec track DeepSeek, Qwen or Claude?
Yes, but only on Enterprise and on the agency Scale and Comprehensive plans. None of them can be bought as a self-serve add-on.
The Enterprise list adds seven engines to the base six and Perplexity: "Claude Sonnet 4 (API)", "GPT 5 Search (API)", "Grok (API)", "Deepseek (API)", "Qwen (API)", "Mistral (API)" and "Meta Spark (API)". Enterprise pricing is custom (Peec pricing).
For agencies, Peec sells credit-based tiers. One credit covers one prompt on one engine for one day (Peec docs). The Essential and Growth tiers list the base six engines. The Scale tier, from $795 a month, lists the same "(API)" engines as Enterprise. Peec's changelog states the boundary: "API model selection is available on Scale and above" (changelog, 3 August 2026).
What does the "(API)" label tell you about the data?
The public pages do not say. Peec's pages describe its collection method in two ways that they do not reconcile.
Peec's introductory docs describe a browser-based method. A section headed "Why we don't use APIs" says: "Instead of APIs, Peec AI uses browser automation to interact with AI platforms through their web interfaces." It adds that "Peec AI doesn't depend on official APIs" (Peec docs).
Other Peec pages refer to API-based surfaces. The pricing tables mark seven engines "(API)". The changelog speaks of "API model selection". The model channels docs give "Perplexity API" as an example surface and mention "API-only models" (model channels docs). None of these pages explains how the "(API)" engines fit the browser-based method.
The distinction matters because two kinds of tracking measure different things:
- Consumer answer-engine monitoring records what a person sees when they type a prompt into an app such as ChatGPT or Gemini, including any web search the app runs.
- API-based model testing sends the prompt to the model through its developer API. The answer depends on settings the tool chooses, such as the model version and whether web search is on.
If the "(API)" engines are queried through developer APIs, their results show how the model answers. They would not show what a user of the DeepSeek, Qwen or Claude app sees. Peec's public docs do not describe settings for these runs.
If a project includes both kinds of engine, check whether the dashboard reports them together or separately. Our engine coverage overview shows how other tools handle the same split.
Which Claude model, and how current is the list?
The pricing pages name one Claude model: "Claude Sonnet 4 (API)". Peec's changelog names newer ones. Its 13 October 2026 entry mentions tracking "Claude Fable and Claude Opus".
The same entry says fan-out queries, the web searches a model runs before it answers, now include searches from "Claude Haiku 4.5 and Claude Sonnet 5" (changelog).
The changelog does not say which plans the newer Claude models are on. It also does not say whether Claude Haiku 4.5 is a tracked engine or appears only in fan-out data.
A smaller mismatch sits on the pricing page itself. The Enterprise card says "Up to 13 LLM models tracked", while the Enterprise table lists 14 engine names.
Peec's docs also note that the model behind a named surface can change while the name stays the same (model channels docs). A plan table is a snapshot of a moving list.
How the plans fit different workflows
Self-serve Peec fits teams whose buyers mainly use ChatGPT, Google's AI surfaces, Copilot, Gemini or Naver. Three engines are included, chosen per project. Perplexity, or a fourth engine, is a priced extra that grows with plan size.
Teams that need DeepSeek, Qwen or Claude face a different trade-off. Those engines require Enterprise or agency Scale and above, and they carry an "(API)" label that Peec's method docs do not explain. The public pages also do not settle which Claude model runs today.
Cost per answer is covered in Peec AI vs Otterly.AI, and reporting differences in Peec AI vs Profound.