Model catalog
A panel of rivals, not a single oracle
Every claim you check is voted on by up to 4 frontier models from different providers, then an independent judge weighs their reasoning into one verdict. These are the models you can put on your panel — each adds its credit weight to the per-claim price, so the quote always reflects the panel you chose.
| Model | Provider | Credit weight / claim | |
|---|---|---|---|
| Claude Opus 4.8 | Anthropic | 8 | |
| Claude Sonnet 4.6 | Anthropic | 5 | Recommended panel |
| GPT-5.5 | OpenAI | 10 | |
| GPT-5.4 Mini | OpenAI | 2 | Recommended panel |
| Gemini 3.5 Flash | 3 | Recommended panel | |
| Gemini 3.1 Pro | 4 | ||
| DeepSeek V4 Flash | DeepSeek | 1 | Recommended panel |
| Grok 4.3 | xAI | 1 | |
| Kimi K2.6 | Moonshot AI | 2 | |
| GLM 5 | Z.ai | 1 | |
| Qwen3.7 Max | Qwen | 2 | |
| Llama 4 Maverick | Meta | 1 | |
| Mistral Medium 3.1 | Mistral | 1 |
The lineup evolves as new models ship — this page always reflects the current catalog.
How the panel works
- Panel votes. Each model on your panel checks every claim independently — Model Fact-Check from its own knowledge, Live Research with live web search and citations.
- An independent judge. Claude Opus 4.8 synthesizes the panel's votes, reasoning, and evidence into one consensus verdict with a calibrated confidence score. You can pick a different judge for your organization.
- An adversarial challenger. When the panel is unanimous and confident, a cheap independent model tries to refute the verdict before it's confirmed — dissent escalates to the full judge.
- Spell & grammar (Tier 1) runs on a fixed fast model (Gemini 3.5 Flash) and isn't part of the picker.
Read the full pipeline write-up on the methodology page, or see pricing for how credit weights turn into a quote.
Customize your panel
Signed-in organization admins choose which models vote and which judge presides on the panel settings page. New accounts start on the recommended panel above.