New in the DEJAN app: Gemini 4 Argon
Asking Argon
- Find your brand, your competitors and your topics
- Ask Argon which brands it names for each topic
- Ask Argon what it links to your brand
- Ask Argon which brands it names for search questions
- Ask Argon how relevant your brand is to each topic
- Ask Gemini 3.8 Flash the same questions
- Argon compares its answers with Flash
What Gemini 4 Argon says about IKEA
Gemini 4 Argon named IKEA in 59 of 165 answers at an average position of 1.4 and rated it relevant in 94% of answers about its topics. Gemini 3.8 Flash named it in 55 of 165 on the same questions.
Gemini 4 Argon Gemini 3.8 Flash (Flash figures appear when its answers are in)
Where each model places your brand
Brands named for your topics
Share of all brand mentions when each model was asked about your topics.
Brands named for search-style questions
Questions phrased the way people search, answered from each model's own knowledge.
Competitors in Argon's answers
Mentions across topic answers and search-style questions, with where each brand appears in Argon's lists.
Your topics
How often each model names your brand for a topic, and how relevant it rates your brand when asked directly.
Search-style questions
Your brand's position in each model's answer. A dash means the model did not name your brand.
What Argon links to your brand
Larger terms came up more often. Outlined terms came up in both models.
What 3.8 Flash links to your brand
The same question, asked of Gemini 3.8 Flash.
What stands out in Argon's answers
- IKEA is the most named brand: 59 mentions, next is Mio with 21.
- Strongest topic: Bedroom furniture, where AI names IKEA in 100% of answers.
- When AI names IKEA, it lists it at position 1.4 on average.
Gemini 4 Argon compared with Gemini 3.8 Flash
Gemini 3.8 Flash is answering the same questions. The comparison appears here when it is ready.
| Same questions, two models | Argon | 3.8 Flash |
|---|
Brands only Argon names
Brands only 3.8 Flash names
What differs
Both models agree that 0 competitors are named more often than IKEA, and both score IKEA at 100.0% relevance across nine core furniture topics.
- Share of mentions: Gemini 4 Argon gives IKEA 10.4% (and 10.0% in top brands), up from 9.6% (and 9.4% in top brands) in Gemini 3.8 Flash.
- Recall: Unprompted mentions rose from 94.0% in Flash to 100.0% in Argon. Search-style answers rose from 80.0% in Flash to 90.0% in Argon.
- Relevance and position: Overall relevance dropped from 100.0% in Flash to 94.0% in Argon. Average position slipped from 1.1 in Flash to 1.4 in Argon.
What is new with Argon
- Topic shift: Relevance for "Sustainable home furnishings" fell from 100.0% in Flash to 40.0% in Argon.
- Competitors: Argon adds Mio (3.4%), String (3.0%), Svenskt Tenn (2.0%), Swedese (2.0%), Artek (1.8%), and Fritz Hansen (1.6%). Flash included Muuto (2.8%), West Elm (1.8%), Pottery Barn (1.8%), String Furniture (1.8%), Georg Jensen (1.6%), and BoConcept (1.6%).
- Brand terms: Argon uniquely links "Älmhult", "Blue Frakta bag", and "Self-assembly", dropping Flash terms like "Cinnamon buns" and "Affordable home decor".
What it means for the brand
Google's newer model mentions IKEA more often, though slightly lower down the page. IKEA should review its content on "Sustainable home furnishings" due to the large drop in Argon, and watch local competitors like Mio, String, and Svenskt Tenn, which Argon now favors.
Written by Gemini 4 Argon from the figures on this page.
Open this analysis in the DEJAN app
Your project keeps these answers and adds OpenAI and Anthropic, so you can compare all three platforms. New accounts get $10 of starter credit, enough to set up your first project. No card required.
Ask Argon about your own brand
Enter your website at the top of this page. Argon and Gemini 3.8 Flash answer the same questions about your brand, free and without sign-in.
Argon now steers Content Optimizer
Content Optimizer tests edits to your page against an AI ranker and keeps the edits that move it higher. When a run stops improving, Recommend Steering sends the run's results, the competitors' advantages and the earlier rounds to an expert model, which writes directions for the next rounds.
Gemini 4 Argon is now the Google expert model for Recommend Steering. It runs at a high thinking level. Claude Opus 5.5 stays the Anthropic expert model.
| Model | gemini-4-argon, stable |
|---|---|
| Context window | 1,048,576 tokens |
| Output per reply | 65,536 tokens |
| Thinking levels | Low, medium, high |
| Search grounding | Supported |