AI Visibility Benchmark
Atopic Dermatitis Market, Austria
This analysis looks at the variation in AI Visibility when the same question is asked using ChatGPT, Claude, Gemini and Perplexity. We measured AI Visibility of therapy classes, molecules, brands and mechanisms of action in the atopic dermatitis market in Austria.
Two therapy classes, JAK inhibitors and biologics, have high AI Visibility across all AI models. AI Visibility of molecules, brands and mechanisms of action shows that AI representation can be very different across AI models.
- stakeholder
- Dermatologists
- language
- German
- total mentions
- 12,829
- AI systems
- 4
- measured
- 24 September 2026
AI Visibility % by AI model
The heatmap below shows the % of responses that mention the class, molecule, brand or MoA at least once.
N = 752 responses · 12,829 total mentions
| Claude | ChatGPT | Perplexity | Gemini | |
|---|---|---|---|---|
| Select any entity to explore | ||||
| Therapy classesJAK inhibitors: 87–99% · Biologics: 80–99% | ||||
| JAK Inhibitors | 87% | 89% | 91% | 99% |
| Biologics | 80% | 82% | 81% | 99% |
| MoleculesLebrikizumab: 38% in Perplexity · 94% in Gemini | ||||
| Dupilumab | 88% | 92% | 86% | 99% |
| Upadacitinib | 71% | 80% | 67% | 99% |
| Tralokinumab | 55% | 77% | 78% | 97% |
| Abrocitinib | 61% | 77% | 63% | 98% |
| Baricitinib | 51% | 49% | 51% | 91% |
| Lebrikizumab | 38% | 56% | 38% | 94% |
| Nemolizumab | 22% | 16% | 7% | 61% |
| BrandsDupixent: 14% in Claude · 61% in Gemini | ||||
| Dupixent | 14% | 36% | 22% | 61% |
| Adtralza | 4% | 9% | 3% | 62% |
| Rinvoq | 2% | 5% | 3% | 44% |
| Cibinqo | 2% | 4% | 1% | 45% |
| Nemluvio | 1% | 3% | <1% | 40% |
| Ebglyss | 6% | 5% | 1% | 30% |
| Olumiant | 1% | 2% | <1% | 39% |
| MechanismsIL-13: 4% in Perplexity · 85% in Gemini | ||||
| IL-4Rα | 29% | 20% | 12% | 87% |
| IL-13 | 22% | 12% | 4% | 85% |
| IL-31RA | 8% | <1% | <1% | 53% |
| JAK1 | 14% | 1% | <1% | 43% |
| JAK1/JAK2 | 0% | <1% | 0% | 15% |
02 · Competitive position
Dupilumab leads everywhere. AI models are not aligned on #2.
- Claude
- ChatGPT
- Perplexity
- Gemini
- Dupilumab86–99%
- Upadacitinib67–99%
- Tralokinumab55–97%
- Abrocitinib61–98%
- Baricitinib49–91%
- Lebrikizumab38–94%
- Nemolizumab7–61%
A therapy's position in AI depends on which system is asked.
03 · Molecule and brand
The molecule is named. The brand, far less often.
For every medicine and in every system, the brand name appears in fewer answers than the molecule. Compared to other AI models, Gemini uses brand names far more often.
When AI discusses the molecule, how often does it also name the brand?
Of the answers naming the molecule, the share that also name its brand.
| Molecule and brand | Claude | ChatGPT | Perplexity | Gemini |
|---|---|---|---|---|
| DupilumabDupixent | 16% | 39% | 25% | 61% |
| UpadacitinibRinvoq | 2% | 5% | 4% | 44% |
| TralokinumabAdtralza | 5% | 11% | 3% | 62% |
| AbrocitinibCibinqo | 3% | 3% | 2% | 46% |
| BaricitinibOlumiant | 2% | 3% | 1% | 42% |
| LebrikizumabEbglyss | 17% | 8% | 3% | 32% |
| NemolizumabNemluvio | 5% | 19% | 0% | 62% |
04 · What Visibility can't show
Visibility counts what appears. It doesn't tell you what was said.
JAK inhibitors are often named as a class, without a JAK molecule or brand name.
1 in 4
In Perplexity, 42 of 171 answers that mention JAK inhibitors do not name a JAK molecule or brand name.
Theralexis drives your molecule ↔ class association.
Visible under the US brand name.
55 of 145
55 responses used ‘Adbry’, the US brand name, instead of ‘Adtralza’.
Theralexis guides correct brand naming for your country.
Gemini talks about mechanism of action more than any other AI model.
53% vs <1%
‘IL-31RA’ appears in 53% of Gemini's answers, compared with 8% in Claude and less than 1% in both ChatGPT and Perplexity.
Theralexis shapes MoA narratives across AI models.
Beyond Visibility
Visibility tells you whether you show up. AI Representation tells you how AI models talk about you.
This AI Visibility analysis illustrates who is mentioned across AI models. Visibility is a hygiene factor, but it is not enough to differentiate your brand. What matters is how AI models represent your products — and whether that representation aligns with your business objectives and strategic priorities.
This is where Theralexis comes in. Our experts are experienced pharma commercial leaders who help pharma marketing and omnichannel teams identify and focus on the biggest commercial opportunities. We use an AI Representation assessment as the starting point, then provide channel and content recommendations to strengthen how your products are represented across AI.
Your market
See how AI represents your therapy, not just whether it appears.
We'll show you how each AI system describes your therapy and its competitors to dermatologists, what is shaping that picture, and where content can change it.
About this benchmark
What was measured
Four AI systems each answered the same questions about atopic dermatitis, asked in German from the perspective of dermatologists in Austria.
Visibility is the share of answers that mention a therapy class, molecule, brand or mechanism at least once. Share of Voice, shown in the data tables, is a term's share of total mentions within its own type (e.g. all mentions of Rinvoq compared with all other brand mentions).
Mechanism Visibility measures whether the terminology appears. This analysis does not show which therapy a mechanism was linked to, or whether it was described correctly. That would be part of a broader Theralexis analysis.
What was not measured
This is a Visibility benchmark. It measures which therapy classes, molecules, brands and mechanisms AI systems mention, and how often. It does not show how those systems talk about a therapy, whether it is described accurately, which attributes it is associated with, or whether that representation aligns with a brand's strategy.
A product with high AI Visibility might still be working against your commercial objectives. That's why AI Visibility is not necessarily a competitive differentiator. When AI talks about your brand, you need to understand what is said.
This is why a full Theralexis assessment goes much further — beyond AI Visibility metrics provided by generic SEO and GEO tools. We reveal how AI models talk about your product, how that aligns with your strategic priorities, the sources and channels shaping it, and what actions can strengthen your competitive position.
- Market
- Atopic Dermatitis · Austria
- Stakeholder
- Dermatologists
- Language
- German
- Responses
- 752
- Measured
- 24 September 2026
- Tracked
- 2 therapy classes · 7 molecules · 7 brands · 5 mechanisms
- AI systems measured
- ChatGPT, Claude, Gemini, Perplexity
Full benchmark data
Every tracked term, by type
Visibility for every tracked term, across all four AI systems and in each. Share of Voice is a term's share of all mentions within its own type; types are never mixed.
Therapy classes
| Term | All systems | Claude | ChatGPT | Perplexity | Gemini | Share of Voice |
|---|---|---|---|---|---|---|
| JAK Inhibitors | 91.6%689 of 752 | 87.2% | 89.4% | 91.0% | 98.9% | 57.2% |
| Biologics | 85.5%643 of 752 | 80.3% | 81.9% | 80.9% | 98.9% | 42.8% |
Molecules
| Term | All systems | Claude | ChatGPT | Perplexity | Gemini | Share of Voice |
|---|---|---|---|---|---|---|
| Dupilumab | 91.5%688 of 752 | 88.3% | 92.0% | 86.2% | 99.5% | 27.3% |
| Upadacitinib | 79.3%596 of 752 | 71.3% | 79.8% | 67.0% | 98.9% | 17.1% |
| Tralokinumab | 76.6%576 of 752 | 55.3% | 76.6% | 77.7% | 96.8% | 14.9% |
| Abrocitinib | 74.7%562 of 752 | 60.6% | 76.6% | 63.3% | 98.4% | 14.3% |
| Baricitinib | 60.4%454 of 752 | 50.5% | 49.5% | 50.5% | 91.0% | 8.7% |
| Lebrikizumab | 56.5%425 of 752 | 37.8% | 56.4% | 37.8% | 94.1% | 12.2% |
| Nemolizumab | 26.7%201 of 752 | 22.3% | 16.5% | 6.9% | 61.2% | 5.6% |
Brands
| Term | All systems | Claude | ChatGPT | Perplexity | Gemini | Share of Voice |
|---|---|---|---|---|---|---|
| Dupixent | 33.1%249 of 752 | 14.4% | 35.6% | 21.8% | 60.6% | 28.6% |
| Adtralza* | 19.3%145 of 752 | 3.7% | 8.5% | 2.7% | 62.2% | 16.7% |
| Rinvoq | 13.2%99 of 752 | 1.6% | 4.8% | 2.7% | 43.6% | 10.8% |
| Cibinqo | 12.9%97 of 752 | 1.6% | 3.7% | 1.1% | 45.2% | 10.3% |
| Nemluvio | 11.3%85 of 752 | 1.1% | 3.2% | 0.5% | 40.4% | 13.3% |
| Ebglyss | 10.6%80 of 752 | 6.4% | 4.8% | 1.1% | 30.3% | 11.4% |
| Olumiant | 10.6%80 of 752 | 1.1% | 1.6% | 0.5% | 39.4% | 9.0% |
* Includes answers using ‘Adbry’, an alias of the tracked Adtralza brand asset.
Mechanisms
Mechanism Visibility shows that the terminology appears. It does not show which therapy it was linked to.
| Term | All systems | Claude | ChatGPT | Perplexity | Gemini | Share of Voice |
|---|---|---|---|---|---|---|
| IL-4Rα | 37.0%278 of 752 | 29.3% | 19.7% | 12.2% | 86.7% | 28.4% |
| IL-13 | 30.7%231 of 752 | 21.8% | 11.7% | 4.3% | 85.1% | 39.5% |
| IL-31RA | 15.6%117 of 752 | 8.0% | 0.5% | 0.5% | 53.2% | 14.6% |
| JAK1 | 14.8%111 of 752 | 14.4% | 1.1% | 0.5% | 43.1% | 15.0% |
| JAK1/JAK2 | 4.0%30 of 752 | 0.0% | 0.5% | 0.0% | 15.4% | 2.5% |