AI Visibility & Prominence · Metastatic colorectal cancer · Spain

AI Visibility and Prominence in Metastatic Colorectal Cancer

Being named is not the same as being named first.

We asked ChatGPT, Claude, Gemini and Perplexity the same 239 questions about treating metastatic colorectal cancer — in Spanish, from the perspective of an oncologist in Spain — and measured which of 22 treatments each system named, and which it named first, across 956 answers.

The clinical question changes the picture: panitumumab is named in 39% of all answers, and in 88% of answers to RAS wild-type questions.

The AI systems agree on the top three. Below them they differ: fruquintinib is sixth in ChatGPT and tenth in Perplexity.

Illustration of a colon with a tumour, between antibody and molecule structures on one side and streams of light flowing into floating AI interface panels on the other.
perspective
Oncologists
language
Spanish
questions
239
answers
956
AI systems
4
measured
20 October 2026

01 · AI Visibility

Bevacizumab is named in 62% of answers, cetuximab in 53%.

AI Visibility of the 22 scored molecules, across all 956 answers or one system's 239.

How many answers name the molecule at least once?The active substance

The 22 scored molecules, with the answers naming each at least once, overall and within each AI system. 956 responses scored, 239 per AI system.
TreatmentTracked asLinkedAI VisibilityClaudeChatGPTPerplexityGemini
1BevacizumabMolecule—62%593 / 95655%13260%14371%17062%148
2CetuximabMolecule—53%504 / 95642%10044%10567%15959%140
3PanitumumabMolecule—39%375 / 95627%6535%8458%13837%88
4RegorafenibMolecule—33%319 / 95623%5435%8342%10134%81
5Trifluridine/​tipiracilMolecule—32%307 / 95619%4535%8342%10033%79
6PembrolizumabMolecule—25%241 / 95615%3717%4044%10624%58
7EncorafenibMolecule—23%222 / 95614%3316%3834%8129%70
8NivolumabMolecule—23%221 / 95615%3714%3341%9723%54
9FruquintinibMolecule—22%210 / 95616%3823%5629%6920%47
10IpilimumabMolecule—20%193 / 95612%2913%3037%8819%46
11AfliberceptMolecule—11%104 / 9569.2%228.4%2016%3810%24
12RamucirumabMolecule—10%96 / 9567.1%177.9%1917%408.4%20
13Adagrasib †Molecule—8.3%79 / 9567.5%187.1%1713%325.0%12
14Trastuzumab †Molecule—7.6%73 / 9565.0%126.7%1612%287.1%17
15Sotorasib †Molecule—7.3%70 / 9567.5%185.9%1411%265.0%12
16Tucatinib †Molecule—6.3%60 / 9564.2%105.4%138.8%216.7%16
17Trastuzumab deruxtecan ‡Molecule—5.2%50 / 9563.8%94.2%106.7%166.3%15
18Binimetinib †Molecule—3.1%30 / 9562.9%73.3%84.2%102.1%5
19Pertuzumab †Molecule—2.5%24 / 9563.3%82.5%63.8%90.4%1
20Larotrectinib ‡Molecule—2.2%21 / 9561.7%41.7%42.5%62.9%7
21DivarasibMolecule—0.4%4 / 9561.3%30.0%00.4%10.0%0
22DaraxonrasibMolecule—0.1%1 / 9560.4%10.0%00.0%00.0%0

Bars are scaled to the highest value in the current view. Every figure is also written out.

Notes on the ranking

  • † No EU authorisation specifically for metastatic colorectal cancer (October 2026). Adagrasib, Trastuzumab, Sotorasib, Tucatinib, Binimetinib, Pertuzumab.
  • ‡ Tumour-agnostic EU authorisation only (biomarker-defined solid tumours); no colorectal-specific indication. Trastuzumab deruxtecan, Larotrectinib.

European Medicines Agency product pages (EPAR), checked 20 October 2026.

02 · Named and named first

Named is not the same as named first.

AI Visibility shows whether a treatment is named. Named first shows whether it is the first of the tracked treatments an answer names.

Later-line oral agents: named at similar rates, named first very differently.

In the 196 answers to later-line questions, regorafenib is named in 76% and trifluridine/tipiracil in 72%. Trifluridine/tipiracil is the first of the three oral agents named in 50% of those answers, regorafenib in 19% and fruquintinib in 12%.

  • AI Visibility (named)
  • Named first among the three oral agents

  1. Regorafenib149 and 38 of 196 answers

  2. Trifluridine/tipiracil142 and 98 of 196 answers

  3. Fruquintinib106 and 23 of 196 answers

Both bars are shares of the same 196 answers, all four AI systems. Named first counts only the three oral agents; answers naming none of them are in the base.

A ranking by AI Visibility puts regorafenib ahead. The order of mention does not: trifluridine/tipiracil comes first in 98 answers, regorafenib in 38.

Across all questions

Panitumumab is the third most named molecule, in 39% of answers, and eighth by named first, at 4%. Trifluridine/tipiracil is fifth by AI Visibility and second by named first.

  • AI Visibility · 956 answers
  • Named first among the 22 molecules · same answers

  1. Bevacizumab593 and 260 of 956 answers · #1 → #1

  2. Cetuximab504 and 100 of 956 answers · #2 → #3 · partner of encorafenib

  3. Panitumumab375 and 37 of 956 answers · #3 → #8

  4. Regorafenib319 and 48 of 956 answers · #4 → #6

  5. Trifluridine/tipiracil307 and 126 of 956 answers · #5 → #2

Show the other molecules

  1. Pembrolizumab241 and 90 of 956 answers · #6 → #4

  2. Encorafenib222 and 76 of 956 answers · #7 → #5 · partner of cetuximab

  3. Nivolumab221 and 38 of 956 answers · #8 → #7 · partner of ipilimumab

  4. Fruquintinib210 and 33 of 956 answers · #9 → #9

  5. Ipilimumab193 and 1 of 956 answers · #10 → #18 · partner of nivolumab

  • AI Visibility · 956 answers
  • Named first among the 22 molecules · same answers
  1. Bevacizumab62% · 27%#1 → #1
  2. Cetuximab53% · 10%#2 → #3 · partner of encorafenib
  3. Panitumumab39% · 4%#3 → #8
  4. Regorafenib33% · 5%#4 → #6
  5. Trifluridine/tipiracil32% · 13%#5 → #2
  6. Pembrolizumab25% · 9%#6 → #4
  7. Encorafenib23% · 8%#7 → #5 · partner of cetuximab
  8. Nivolumab23% · 4%#8 → #7 · partner of ipilimumab
  9. Fruquintinib22% · 3%#9 → #9
  10. Ipilimumab20% · <1%#10 → #18 · partner of nivolumab

Encorafenib and cetuximab (BRAF V600E), and nivolumab and ipilimumab (MSI-H/dMMR), are regimen partners: the order between partners mostly reflects how the regimen is written, so they are not compared with each other on named first.

Order of mention is not a preference, a recommendation or a clinical priority.

03 · Clinical settings

Are you leading where it matters most?

AI Visibility within each clinical setting: the share of that setting's answers, across all four AI systems, that name each relevant treatment.

View
  • =n shared rank
  • Bracket — regimen partners
  • † No EU authorisation specifically for metastatic colorectal cancer (October 2026).

First-line · 100 answers

  1. 1Bevacizumab85%
  2. 2Cetuximab65%
  3. 3Panitumumab46%

RAS wild-type · 72 answers

  1. =1Cetuximab88%
  2. =1Panitumumab88%
  3. 3Bevacizumab83%

BRAF V600E · 56 answers

  1. =1Cetuximab100%
  2. =1Encorafenib100%
  3. 3Bevacizumab59%
  4. 4Binimetinib†48%

Later-line · 196 answers

Oral agents

  1. 1Regorafenib76%
  2. 2Trifluridine/tipiracil72%
  3. 3Fruquintinib54%

Biologics / combination regimens

  1. 1Bevacizumab74%
  2. 2Cetuximab44%
  3. 3Panitumumab27%
  4. 4Aflibercept23%
  5. 5Ramucirumab22%

MSI-H/dMMR · 72 answers

  1. =1Nivolumab96%
  2. =1Ipilimumab96%
  3. 3Pembrolizumab94%

Each treatment's AI Visibility across all 956 answers, and within the questions about the setting it is used in.

  • All questions · 956 answers
  • Questions about its setting
  1. RAS wild-type · 72 answers
  2. Bevacizumab62% → 83%Named first 24%
  3. Cetuximab53% → 88%Named first 44%
  4. Panitumumab39% → 88%Named first 26%
  5. Later-line · 196 answers
  6. Regorafenib33% → 76%Named first 12%
  7. Trifluridine/tipiracil32% → 72%Named first 37%
  8. Fruquintinib22% → 54%Named first 10%
  9. MSI-H/dMMR · 72 answers
  10. Pembrolizumab25% → 94%Named first 54%
  11. Nivolumab23% → 96%Named first 39% · partner of ipilimumab
  12. Ipilimumab20% → 96%Named first 0% · partner of nivolumab
  13. BRAF V600E · 56 answers
  14. Encorafenib23% → 100%Named first 95% · partner of cetuximab

Bevacizumab is shown in RAS wild-type questions, where it is compared with anti-EGFR therapy. In first-line questions it is named in 85%.

Named first counts the first of all scored molecules named in the setting's answers. Between regimen partners the order mostly reflects how the regimen is written, so partners are not compared with each other on it.

  • Each panel is one set of questions with its own base. Treatments are compared within a panel, never across panels, and are not interchangeable.
  • A panel lists the scored treatments relevant to its setting. A treatment that is not listed is not relevant there; it is not counted as zero. Unscored agents, such as chemotherapy, are not shown.
  • Settings are shown only with 40 or more answers. Being named is not a preference.

04 · Across AI systems

The top three are the same in every AI system. Below them, the order changes.

Each molecule's rank by AI Visibility within each system's 239 answers, for the most-named molecules. Tied molecules share a rank. Every molecule's figure in each system is in the AI Visibility ranking.

  • Claude
  • ChatGPT
  • Perplexity
  • Gemini
  1. Bevacizumab1
  2. Cetuximab2
  3. Panitumumab3
  4. Regorafenib4–5
  5. Trifluridine/tipiracil4–6
  6. Pembrolizumab4–7fourth in Perplexity, seventh in ChatGPT and Gemini
  7. Encorafenib6–9sixth in Gemini, ninth in Claude and Perplexity
  8. Nivolumab7–9
  9. Fruquintinib6–10sixth in ChatGPT and Claude, tenth in Perplexity
  10. Ipilimumab8–10
Scored molecules named per answer
  • Claude2.9
  • ChatGPT3.4
  • Perplexity5.6
  • Gemini3.9

Perplexity names 5.6 scored molecules per answer on average, Claude 2.9. A system that lists more treatments raises every molecule's AI Visibility at once — so compare ranks within a system, not levels across systems.

Measuring one AI system describes one AI system.

05 · Question categories

Which treatments are named depends on what is asked.

Share of answers in each question category naming each of the most-named molecules, with its rank among them. Each column compares the molecules on the same questions; the columns differ in question mix.

  • Rank moves three or more places from all questions
  • #n rank among these molecules
  • =n shared rank
All questions956 answersEfficacy404 answersSafety352 answersConvenience140 answersAdherence60 answers
Bevacizumab#1#1#1#1#3
Cetuximab#2#2#2=2#5
Panitumumab#3#3#3#5=6
Regorafenib#4#7#4=2#2
Trifluridine/tipiracil#5#8#5#4#1
Pembrolizumab#6#5#6#8=6
Encorafenib#7#4#10#7=6
Nivolumab#8#6#7#9=6
Fruquintinib#9#10#8#6#4
Ipilimumab#10#9#9#10=6

All questions · 956 answersEfficacy · 404 answers

  1. Bevacizumab62% #1#1
  2. Cetuximab53% #2#2
  3. Panitumumab39% #3#3
  4. Regorafenib33% #4#7
  5. Trifluridine/​tipiracil32% #5#8
  6. Pembrolizumab25% #6#5
  7. Encorafenib23% #7#4
  8. Nivolumab23% #8#6
  9. Fruquintinib22% #9#10
  10. Ipilimumab20% #10#9

All questions · 956 answersSafety · 352 answers

  1. Bevacizumab62% #1#1
  2. Cetuximab53% #2#2
  3. Panitumumab39% #3#3
  4. Regorafenib33% #4#4
  5. Trifluridine/​tipiracil32% #5#5
  6. Pembrolizumab25% #6#6
  7. Encorafenib23% #7#10
  8. Nivolumab23% #8#7
  9. Fruquintinib22% #9#8
  10. Ipilimumab20% #10#9

All questions · 956 answersConvenience · 140 answers

  1. Bevacizumab62% #1#1
  2. Cetuximab53% #2=2
  3. Panitumumab39% #3#5
  4. Regorafenib33% #4=2
  5. Trifluridine/​tipiracil32% #5#4
  6. Pembrolizumab25% #6#8
  7. Encorafenib23% #7#7
  8. Nivolumab23% #8#9
  9. Fruquintinib22% #9#6
  10. Ipilimumab20% #10#10

All questions · 956 answersAdherence · 60 answers

  1. Bevacizumab62% #1#3
  2. Cetuximab53% #2#5
  3. Panitumumab39% #3=6
  4. Regorafenib33% #4#2
  5. Trifluridine/​tipiracil32% #5#1
  6. Pembrolizumab25% #6=6
  7. Encorafenib23% #7=6
  8. Nivolumab23% #8=6
  9. Fruquintinib22% #9#4
  10. Ipilimumab20% #10=6
  • 9 of the 15 adherence questions concern oral therapies. Of the questions stating BRAF V600E, 9 are about efficacy and 4 about safety.
  • Being named in answers to efficacy or safety questions is not a measure of efficacy or safety.

Full assessment

What a full assessment reveals

Example: Your brandSpain · Oncologists

  • Competitive positioning

    Your brand vs competitorsYour brandCompetitor ACompetitor BCompetitor C

  • Product attributes

    EfficacySafetyDosingOutcomesYour brandCompetitor ACompetitor B

  • Strategic alignment

    Intended vs describedIntended positioningPriority messageDifferentiationTarget role

  • Source intelligence

    Sources shaping the answerGuidelinesPublicationsPharma-ownedHCP / professional

  • Content recommendations

    Approved claimsEvidenceExisting contentOpportunities

This public benchmark shows which treatments AI names, and which it names first, across clinical settings, question types and AI systems. A Theralexis assessment goes further: how your brand is positioned against its competitors, what shapes how AI describes it, and where the strategic opportunities are.

Your market

Interested in a full assessment?

Go beyond AI Visibility. Understand how AI represents your brand at key HCP decision moments, how you compare with competitors, and where opportunities exist to strengthen your competitive position.

Schedule a meeting

About this benchmark

What was measured

One measurement, on 20 October 2026: 239 Spanish-language questions, 4 AI systems with web search, one answer per question per system. The results depend on the questions asked and on the systems answering them — another question set, system or date could name and order these treatments differently. AI Visibility and named first are shares of the answers in scope, from two separate measurements of the same answers. They are not measures of clinical preference, recommendation or superiority, and being named says nothing about how a treatment is described.

Definitions, clinical settings and limitations

One measurement: ChatGPT, Claude, Gemini and Perplexity, each with web search, answered the same questions about the treatment of metastatic colorectal cancer — in Spanish, from the perspective of an oncologist in Spain — once each, on 20 October 2026.

AI Visibility — the share of answers in scope that name a molecule at least once.

Named first — the share of answers in scope in which a molecule is the first of the tracked molecules named: all scored molecules, or, for the later-line comparison, the three oral agents. Answers naming none of them stay in the base. AI Visibility and named first come from two separate measurements of the same answers; every figure uses one of them, never both. The two differ on whether a molecule is named in about one mention in a hundred.

Clinical settings — read from each question's own wording: the biomarker it states (RAS wild-type, BRAF V600E, MSI-H/dMMR) or the line of treatment (first-line; later-line, including refractory, pretreated and post-progression disease). A setting lists only the treatments relevant to it and is shown only with 40 or more answers, so no figure for a single AI system is shown inside a setting. Removing, for each treatment, the questions that name it changes no order within a setting.

Regimen partners — encorafenib with cetuximab (BRAF V600E) and nivolumab with ipilimumab (MSI-H/dMMR) are given together. They are bracketed, not compared: the order between partners mostly reflects how the regimen is written. Trifluridine/tipiracil can be given with bevacizumab, which is why the later-line setting ranks oral agents and biologics separately.

Question categories — efficacy, safety, convenience and adherence, every question in exactly one. The columns differ in size and in mix.

What this does not show

Order of mention is not a preference, a recommendation or a clinical priority, and being named says nothing about clinical correctness or about how a treatment is described. The benchmark does not show which sources shaped an answer, or why one treatment comes before another. Unscored agents, including chemotherapy backbones, are not ranked and are not zero. One run in one market, language and audience, with four lighter-tier AI systems; results may differ elsewhere or over time.

Publisher
Theralexis
Market
Metastatic colorectal cancer · Spain
Perspective
Oncologists
Language
Spanish
Questions
239
Answers
956
Measured
Published
Tracked
22 scored molecules
Clinical settings
First-line, RAS wild-type, BRAF V600E, Later-line, MSI-H/dMMR
AI systems measured
ChatGPT, Claude, Gemini, Perplexity

† No EU authorisation specifically for metastatic colorectal cancer (October 2026). Adagrasib, Trastuzumab, Sotorasib, Tucatinib, Binimetinib, Pertuzumab.

‡ Tumour-agnostic EU authorisation only (biomarker-defined solid tumours); no colorectal-specific indication. Trastuzumab deruxtecan, Larotrectinib.

European Medicines Agency product pages (EPAR), checked 20 October 2026.

Full benchmark

Explore the full benchmark.

The full benchmark goes deeper — every treatment by clinical setting, question and AI system, and where the order changes.

Schedule a meeting