No model provider can show you how their competitors talk about your brand. Aeoniti is the only neutral observatory that watches the three engines where buyers ask product questions — ChatGPT, Gemini, and Google AI Overview. Other engines are on the roadmap with no committed dates: we add them when customers tell us their buyers actually use them.
| Engine | Vendor | Model | Via | Notes |
|---|---|---|---|---|
| ChatGPT | OpenAI | openai/gpt-5.4-nano | OpenRouter | Probed via OpenRouter |
| Gemini | google/gemini-3.5-flash | OpenRouter | Probed via OpenRouter | |
| Google AI Overview | perplexity-sonar (AIO proxy) | Perplexity Sonar | Same brand-mention coverage as direct SERP-API at roughly 25% of the cost |
We add engines when customers tell us their buyers actually use them. These are the next candidates — listed for transparency, not as a launch promise.
| Engine | Vendor | Model | Notes |
|---|---|---|---|
| Claude | Anthropic | claude-haiku-3.5 | Subject to API availability + customer demand |
| Grok | xAI | grok-4 | Subject to API availability + customer demand |
| DeepSeek | DeepSeek | deepseek-v3.1 | Subject to API availability + customer demand |
| Llama | Meta | llama-4 | Subject to API availability + customer demand |
Because more engines doesn't mean better data. We probe the three engines where commercial-intent buyers actually ask product questions today. Adding a fourth engine that 0.3% of your buyers use just adds cost — to us, and to the LLM cost cap on your bill. When that changes, we change.
Engine you want covered? Email [email protected] with the model card link and the buyer use-case.
Every new AI engine that ships strengthens Aeoniti's moat without us doing anything. OpenAI will never benchmark itself against Google; Google will never benchmark itself against OpenAI. The third-party neutral position is the only one that gets to compare them — that's our position by structural necessity.
As the AI landscape fragments — and it will, dramatically — the value of a single observatory that watches every assistant grows. Today we cover 3. We add engines on the buyer-evidence trigger, not on the press-release trigger. Single-vendor dashboards can't catch up — they're structurally locked into one model family.