The model landscape
Mentova tracks more than 50 AI models from 14 providers, including OpenAI, Anthropic, Google, Meta, xAI, Mistral, DeepSeek, Perplexity, Cohere, and more. The full registry spans general-purpose chat models, reasoning models, and search-augmented surfaces.
The model landscape changes fast. Mentova keeps the registry up to date so you can track new models as they gain user share, without changing your brand setup.
Model tiers
Every model in the registry belongs to one of three tiers:
| Tier | Description | Slot cost |
|---|---|---|
| Standard | Capable, cost-efficient models - good baseline coverage | 1 slot |
| Premium | Leading frontier models with the broadest user bases | 3 slots |
| Ultra | Highest-capability, highest-cost models | 5 slots |
Popular standard models include ChatGPT, Gemini Flash (latest), Perplexity Sonar, Claude Haiku, and DeepSeek V3. Popular premium models include ChatGPT Premium, GPT-5.4, Claude Sonnet, Gemini 3.1 Pro, and Grok 4.
Default models
Each assistant is represented by a class (ChatGPT, Gemini, Perplexity, Claude…). Unless you change it, a class runs the model chosen to represent that assistant — ChatGPT for the ChatGPT class, Gemini Flash (latest) for Gemini, Sonar for Perplexity.
These defaults are reviewed monthly, and they are not picked on price: we compare candidate models on the brands they actually name, after first measuring how much each model disagrees with itself. Ask one model the same question twice and it does not return the same list — so the gap between two models only means something measured against that gap. A model becomes the default only once the brands it names are indistinguishable from those of the most expensive model in its class.
Pro plans and above can pin a different model inside a class. ChatGPT Premium, OpenAI's top-tier model, is one: roughly 4.4× the price per call, which earns it a single pass per question where the default gets three. Three passes on a common model often beat one pass on an expensive one, precisely because of that disagreement a model has with itself — which is why the default is not the priciest model.
Google Search surfaces
In addition to chat models, Mentova can track Google AI Overviews and Google AI Mode - the AI-generated summaries that now appear at the top of many Google search results. These are tracked via a dedicated SERP integration rather than the standard chat API, and they require a compatible plan.
This matters because Google AI Overviews are seen by a very large audience: they appear directly in Google Search results before users even visit a website.
Selecting models for your brand
You choose which models to track per brand in Settings → Models. A few guidelines:
- Start with "Most Used" models - the models flagged as popular in the registry represent the largest share of actual AI usage today.
- Add "Trending" models - models gaining traction quickly are worth adding early; early visibility data gives you a baseline to measure against as these models grow.
- Match your audience - if your product is aimed at developers, models like Claude and Gemini see heavier usage in that segment. If it is consumer-facing, ChatGPT and Google AI are the priority.
- Watch your credit budget - every premium model you add multiplies campaign cost by 3 slots. See the Credits & Slots page.
Why model coverage matters
No single AI model dominates all use cases or all geographies. A brand might be well-cited by ChatGPT but invisible to Claude or Gemini. Tracking multiple models reveals which parts of the AI landscape you own and where you have blind spots.
Conversely, running every model on every campaign will drain your credits quickly. A good practice is to run broad coverage (many models) monthly, and use targeted campaigns (fewer models) for frequent monitoring.
Common pitfall
Treating all models as equivalent is a mistake. A high Mention Rate on a low-traffic model and a low Mention Rate on GPT-4o represent very different real-world impact. Weight your interpretation by the model's actual user share.