What the Machines Said — San Francisco Fine Dining

§ 01 QUINCE 627 mentions

Quince

627

total mentions

Michael Tusk has been doing the same thing quietly and brilliantly for twenty years. Turns out AI noticed. Quince is what consensus looks like when you stop trying to impress anyone — all four platforms are within 34 mentions of each other, the tightest spread at the top of the dataset.

ChatGPT156 mentions

Solid at 156, below Claude but consistent. Describes Quince with the expected emphasis on prix fixe and produce-driven California cuisine. Correct and unremarkable — which is itself a data point about how much ChatGPT reaches for specificity when describing the city's most-mentioned restaurant.

Claude170 mentions

Leads the four at 170. The platform that recommends Quince most often also spends the most time on Tusk's sourcing ethos, the Michelin three-star credentials, and the connection between the menu and the farms behind it.

Gemini165 mentions

165 — nearly tied with Claude and the most specific of the four about Quince's sourcing. Gemini covers the Sonoma farms, the pasta program, and the Jackson Square building with more editorial texture than ChatGPT manages.

Perplexity136 mentions

The lowest of the four at 136 — a 20% gap from Claude's lead. Still 136 mentions for one restaurant, which is the floor in a dataset where consensus is its own kind of argument. Perplexity surfaces Quince consistently but without the frequency of the others.

§ 02 GARY DANKO 523 mentions

Gary Danko

523

total mentions

The safe, impeccable choice for every corporate dinner and anniversary in San Francisco for twenty years. Perplexity leads with 168 — more than twice ChatGPT's count. Danko is discovery by reputation, and Perplexity is where reputation lives. The 118-mention gap between ChatGPT and Gemini is the widest platform split in the top five.

ChatGPT69 mentions

69 mentions — the lowest of the four by a significant margin. ChatGPT's coverage treats Danko as one among many SF fine dining options rather than the institution it is. The name appears, the accolades are noted, and that's where the description usually ends.

Claude99 mentions

The middle ground at 99 — above ChatGPT but well below Gemini. Claude's Danko references lean on the tasting menu format and the prix fixe structure, correct but rarely specific about what makes a Danko meal distinctive from a decade of similar occasions.

Gemini187 mentions

The dominant voice at 187 — nearly three times ChatGPT's count. Gemini treats Danko as a flagship recommendation and the restaurant it most readily surfaces when asked about San Francisco special occasions. The most enthusiastic platform by a considerable margin.

Perplexity168 mentions

168 and second-highest. Perplexity and Gemini together account for 355 of Danko's 523 mentions — 68% of its total visibility. The pattern confirms that Danko's discovery happens through ranking and aggregation, not editorial discovery.

§ 03 BENU 476 mentions

Benu

476

total mentions

Corey Lee built Benu to last — and apparently so did his press strategy. The most evenly distributed platform profile in the top five: all four within 24 mentions of each other. The first Korean-American chef to hold three Michelin stars registers as consensus, which is its own kind of statement about how the machines process culinary history.

ChatGPT131 mentions

Leads the four at 131. ChatGPT mentions Benu early in responses about Michelin fine dining in SF, consistently noting the three stars and Corey Lee's biography. One of ChatGPT's stronger SF profiles — specific rather than generic.

Claude120 mentions

11 mentions behind ChatGPT at 120. Claude describes Benu's Korean-influenced tasting menu with accurate detail — one of the most consistent restaurant profiles in Claude's SF dataset. The accuracy is there; the recommendation frequency is close behind.

Gemini118 mentions

118 and nearly tied with Claude. Gemini's Benu coverage is substantive — one of the few restaurants where Gemini mentions specific course structures rather than defaulting to "tasting menu." The platform that knows this restaurant well.

Perplexity107 mentions

The lowest at 107 — but a 24-mention range across all four platforms is the tightest spread in the dataset. Benu is the only restaurant in the top ten where all four platforms are within the same order of magnitude. That uniformity is the finding.

§ 04 ATELIER CRENN 458 mentions

Atelier Crenn

458

total mentions

Dominique Crenn has always had a point of view. Turns out AI has one about her too. Claude and Gemini together account for 60% of her total visibility — ChatGPT, with 67 mentions, treats her as a footnote she is not. The gap between ChatGPT's 67 and Claude's 142 is the widest single-platform divergence in the top five.

ChatGPT67 mentions

67 mentions — notably low for a three-Michelin-star restaurant ranked fourth in the city. ChatGPT surfaces Atelier Crenn less frequently than Claude or Gemini by a wide margin, noting the stars and the chef's name without depth.

Claude142 mentions

Leads at 142 and by a considerable margin. Claude's coverage is the most detailed — the "poetic cuisine" framing, the tasting menu as narrative, the plant-forward menu and wine emphasis. The platform that understands Crenn best also recommends her most.

Gemini130 mentions

130 and second to Claude. Gemini surfaces Atelier Crenn with strong awareness of the Michelin context and the sustainability focus — less of the editorial texture that Claude brings, but present and substantive.

Perplexity119 mentions

119 mentions. Perplexity's Crenn references are occasion-driven — anniversary dinners, special events — and accurate in the way that occasion-based recommendations often are: complete enough to get the reservation right, light enough to miss the restaurant entirely.

§ 05 SAISON 397 mentions

Saison

397

total mentions

One of the most expensive meals in America. ChatGPT and Gemini understand that. Claude, with 69 mentions against their combined 245, is still deciding what Saison is. For a restaurant built around open-fire cooking and a fixed tasting format, Claude's hesitation is the most interesting data point in the top five.

ChatGPT120 mentions

Leads at 120, with coverage that emphasizes the open-fire cooking, the price point, and the omakase-style format. One of ChatGPT's stronger profiles in the SF dataset — it reaches for specific details about what Saison actually involves.

Claude69 mentions

69 mentions — the lowest of the four and a sharp contrast with ChatGPT's lead. Claude describes Saison accurately but recommends it less readily, which for one of the most expensive and specific restaurants in America suggests Claude weights different discovery signals.

Gemini125 mentions

Slight lead over ChatGPT at 125. Gemini's Saison coverage emphasizes the wood-fire sourcing and the price — "one of San Francisco's most expensive restaurants" appears frequently — and surfaces it reliably in special-occasion queries.

Perplexity83 mentions

83 mentions, below the other three. Perplexity reaches for Saison in occasion-driven contexts but surfaces it less than Gemini or ChatGPT in open-ended fine dining queries. A platform-specific pattern that's worth tracking.

§ 06 LAZY BEAR 326 mentions

Lazy Bear

326

total mentions

Lazy Bear built its reputation through word of mouth and a ticketing model no one had seen before. Somehow ChatGPT missed the memo. Its 32 mentions represent barely 10% of the total — the sharpest ChatGPT gap in the top ten, in a restaurant where the format itself is the distinguishing feature.

ChatGPT32 mentions

32 mentions. Lazy Bear's model — a ticketed, multi-course dinner in a converted garage — is difficult to categorize by conventional restaurant signals, and ChatGPT appears to struggle with it. The format doesn't fit a standard template, and the count shows it.

Claude70 mentions

70 — more than double ChatGPT but well below Gemini. Claude is familiar with Lazy Bear's concept and recommends it for "unique dining experiences," with fewer mentions of the specific menu or the ticket-model format.

Gemini120 mentions

Leads at 120. Gemini has the strongest representation of Lazy Bear's ticket-dining format and surfaces it with more context about what the experience actually involves — the communal tables, the open kitchen, the ticket-based reservation model.

Perplexity104 mentions

104 and second. Perplexity's coverage sits between Gemini and Claude — it surfaces the restaurant in the right contexts but with less specificity than Gemini about the format. The discovery pattern tracks with editorial awareness more than algorithmic ranking.

§ 07 ACQUERELLO 284 mentions

Acquerello

284

total mentions

A Nob Hill institution with thirty-five years of Italian precision and the kind of loyalty that doesn't require press releases. Perplexity has 38 — less than half of what ChatGPT gives it. The gap suggests a restaurant that earns discovery through traditional means: word of mouth and long-standing reputation rather than editorial platforms.

ChatGPT88 mentions

88 and above Gemini. ChatGPT describes Acquerello with correct detail — the Michelin star, the Italian focus, the anniversary tasting menus and the formality of the room. A solid profile for a restaurant with a three-decade reputation.

Claude60 mentions

60 — below the average. Claude describes Acquerello accurately but less frequently than the restaurant's reputation might suggest. Correct on the essentials, quieter about recommending it in contexts where it should lead.

Gemini98 mentions

Leads at 98, and the only platform that edges out ChatGPT here. Gemini's Acquerello coverage emphasizes the truffle pasta, the wine list, and the Nob Hill setting — more specific than the others about what distinguishes the experience.

Perplexity38 mentions

38 — the lowest by a considerable margin. Perplexity recommends Acquerello roughly one-third as often as Gemini, which for a restaurant with three decades of consistent recognition is a meaningful gap. Discovery here is earned through channels that Perplexity doesn't weight heavily.

§ 08 KOKKARI 253 mentions

Kokkari

253

total mentions

The best Greek restaurant in San Francisco has been hiding in plain sight on Jackson Street for twenty-five years. Gemini and Perplexity account for 85% of its 253 total mentions. Claude has 10. ChatGPT has 25. Both should be embarrassed — and both illustrate how AI visibility can collapse for a specific cuisine type regardless of standing.

ChatGPT25 mentions

25 mentions — low for a restaurant at rank 8. ChatGPT's coverage of Kokkari is minimal. A restaurant open since 1999 on one of the city's best restaurant blocks appears in almost none of ChatGPT's SF fine dining recommendation sets.

Claude10 mentions

10 mentions — the most striking number in the top ten. Claude has essentially not discovered Kokkari. A Greek restaurant open for twenty-five years in a prime Jackson Street location appears in almost none of Claude's SF fine dining recommendations.

Gemini116 mentions

Leads at 116 and by an enormous margin. Gemini accounts for 46% of Kokkari's total visibility — its 116 mentions versus Claude's 10 represents the widest platform gap of any restaurant in the top 20. Gemini is effectively the only platform that has found this restaurant.

Perplexity102 mentions

102 and second. Perplexity and Gemini together account for 85% of Kokkari's total. The discovery pattern for this restaurant is almost entirely driven by two platforms — and absent from the other two in a way that compounds rather than balances.

§ 09 CALIFORNIOS 227 mentions

Californios

227

total mentions

Val Cantú's Mexican fine dining tasting menu was ahead of its time when it opened. Gemini leads at 90 — more than double Claude's count. Platform coverage of this restaurant is uneven, with the lowest-mention platforms producing less confident categorization of what Californios actually is.

ChatGPT43 mentions

43 mentions. ChatGPT describes Californios in the right genre — Mexican fine dining, tasting menu — but without the specificity that Val Cantú's restaurant and its two Michelin stars deserve. Gets the category right; misses the restaurant.

Claude35 mentions

The lowest at 35. A Michelin-starred restaurant with an entirely original concept receives fewer recommendations from Claude than from any other platform. The gap between Claude and Gemini here is meaningful.

Gemini90 mentions

Leads at 90. Gemini surfaces Californios more reliably than the others and is the platform most likely to frame it as a distinctive dining category — Mexican tasting menu as its own genre — rather than folding it into general fine dining.

Perplexity59 mentions

59 — above Claude and ChatGPT but below Gemini. Perplexity's coverage appears mostly in culinary-tourism and occasion-driven contexts, which is consistent with a two-Michelin-star restaurant that earns attention through critical recognition.

§ 10 RICH TABLE 213 mentions

Rich Table

213

total mentions

Evan and Sarah Rich built the definitive Hayes Valley neighborhood restaurant. Perplexity leads with 100 — nearly six times ChatGPT's 18. The algorithms that pull from local editorial know this place. The ones that rely on recognition and marketing don't. A 82-mention gap between Perplexity and ChatGPT is one of the widest in the dataset.

ChatGPT18 mentions

18 mentions — one of the lowest counts for a restaurant of this caliber in the dataset. ChatGPT has essentially not discovered Rich Table, which has been the defining Hayes Valley restaurant for over a decade. The name appears rarely, and never as a lead recommendation.

Claude31 mentions

31 — above ChatGPT but still low. Claude surfaces Rich Table in neighborhood dining contexts more than as a top-of-list recommendation. The platform that's best at word-of-mouth restaurants is still not finding this one frequently enough.

Gemini64 mentions

64 and third. Gemini finds Rich Table with more regularity — the innovative comfort food angle, the Hayes Valley location, the approach to ingredients that earned James Beard nominations. The platform that discovers through culinary vocabulary.

Perplexity100 mentions

Leads at 100 — nearly six times ChatGPT. Perplexity finds Rich Table through local editorial and critical coverage where the restaurant consistently appears. The platform that reads the press finds the restaurants that earn it.

§ 11 BIRDSONG 193 mentions

Birdsong

193

total mentions

Christopher Bleidorn built something genuinely original in SoMa — a restaurant about California produce told through a wood-fired lens. Perplexity found it first, ChatGPT found it last. That order tells you something about who reads the James Beard coverage versus who waits for a restaurant to accumulate mentions.

ChatGPT24 mentions

24 mentions — limited coverage for a restaurant with a James Beard nomination. ChatGPT's Birdsong profile is thin, suggesting the wood-fire, California-produce concept hasn't registered in its recommendation patterns the way the critical record would suggest.

Claude33 mentions

33 — slightly above ChatGPT. Claude surfaces Birdsong with more frequency and tends to mention the SoMa location and the California-focused menu, which is the right framing. Still underweighted relative to the restaurant's reputation.

Gemini58 mentions

58 and third. Gemini's Birdsong coverage tracks the wood-fired cooking style and the seasonal produce focus — the platform that discovers through culinary vocabulary finds this restaurant more than those that don't.

Perplexity78 mentions

Leads at 78. Perplexity finds Birdsong through awards coverage and critical recognition — the James Beard nominations, the local critical record. The discovery pattern here tracks editorial sources more than algorithmic ranking.

§ 12 STATE BIRD PROVISIONS 173 mentions

State Bird Provisions

173

total mentions

Stuart Brioza turned a dim sum cart into a James Beard Award. The platforms that understand the concept best — Claude and Gemini — are nearly even at 47 and 54. ChatGPT's 25 suggests it knows the name but not the idea. Understanding what State Bird actually is appears to be a prerequisite for recommending it with frequency.

ChatGPT25 mentions

25 mentions — low for a James Beard Award winner. ChatGPT knows the name and the dim-sum-style format, but surfaces it less than the Michelin star and the awards recognition would predict. The format doesn't fit a standard template.

Claude47 mentions

47 and nearly even with Perplexity. Claude understands State Bird's concept and recommends it for casual fine dining contexts more than for special occasions — an accurate read of the restaurant's actual positioning.

Gemini54 mentions

Leads at 54. Gemini captures the communal-style service, the family-style ordering, the cart-based format. The platform most equipped to describe what State Bird actually is also recommends it most frequently.

Perplexity47 mentions

47 — even with Claude. Perplexity surfaces State Bird in the "unique San Francisco dining experiences" category and through local editorial. A platform that discovers through awards and critical recognition finds this restaurant reliably.

§ 13 WAYFARE TAVERN 156 mentions

Wayfare Tavern

156

total mentions

Tyler Florence's Financial District institution owns the corporate dinner conversation — and has just 6 ChatGPT mentions to show for it. Perplexity leads at 62, suggesting the discovery happens through itinerary-style queries. ChatGPT is apparently not writing those itineraries for San Francisco's most traditional business-dining destination.

ChatGPT6 mentions

6 mentions — a strikingly low count for a Tyler Florence restaurant that has been a Financial District staple for over a decade. ChatGPT is not writing SF business-dining itineraries that include Wayfare. The name and the chef are known; the recommendation frequency is not.

Claude42 mentions

42 — substantially ahead of ChatGPT. Claude's coverage of Wayfare Tavern tends toward the "American comfort food" framing — the rotisserie chicken, the wood-fired menu, the Financial District location. The right description, delivered more often than ChatGPT manages.

Gemini46 mentions

46 and second. Gemini finds Wayfare through the Financial District location and the Tyler Florence name — two signals the platform weights in SF dining recommendation contexts. Slightly ahead of Claude.

Perplexity62 mentions

Leads at 62. Perplexity's itinerary-style recommendations match well with a destination business-dinner restaurant — the pattern makes sense, and the count reflects it. The platform that aggregates dining recommendations by occasion finds this restaurant reliably.

§ 14 THE PROGRESS 143 mentions

The Progress

143

total mentions

State Bird's quieter sibling. More ambitious, less discovered — by critics and AI alike. ChatGPT's 10 mentions make this one of the most under-indexed restaurants of its caliber in the city. Gemini leads with 55, which is less a compliment than a reminder of how far behind the other three are.

ChatGPT10 mentions

10 mentions for a Michelin-starred State Bird sibling. ChatGPT treats The Progress as an afterthought — a restaurant that earns mention primarily as a satellite to State Bird Provisions rather than on its own terms.

Claude29 mentions

29 mentions. Claude surfaces The Progress in neighborhood dining contexts, sometimes alongside its sibling restaurant. More individual recognition than ChatGPT but still underweighted relative to the restaurant's standing.

Gemini55 mentions

Leads at 55. Gemini is the platform most likely to recommend The Progress on its own terms — as a Fillmore Street restaurant with its own identity — rather than as a State Bird satellite. The platform that understands restaurant context does better here.

Perplexity49 mentions

49 and second. Perplexity finds The Progress through editorial and awards context — the Michelin star, the critical recognition. More individual discovery here than through the State Bird association that drives the others.

§ 15 NOPA 139 mentions

Nopa

139

total mentions

A Mission institution that's fed more off-duty chefs than anywhere in the city for fifteen years. Claude leads with 49, suggesting the platforms that reward longevity and word-of-mouth found it. ChatGPT's 10 mentions suggest it hasn't eaten here yet — which is exactly the kind of restaurant ChatGPT misses.

ChatGPT10 mentions

10 mentions — minimal coverage for a restaurant that has anchored the NoPa neighborhood for fifteen years. ChatGPT misses Nopa almost entirely in fine dining recommendation contexts. The type of restaurant that earns loyalty without press releases doesn't surface here.

Claude49 mentions

Leads at 49. Claude recommends Nopa with more regularity than any other platform — a pattern that tracks with the restaurant's word-of-mouth reputation and its standing in the chef community. The platform that discovers through local knowledge finds this restaurant.

Gemini37 mentions

37 mentions. Gemini's Nopa coverage surfaces in late-night dining and neighborhood contexts — the late kitchen hours, the NoPa designation. Correct positioning, moderate frequency.

Perplexity43 mentions

43 — second. Perplexity finds Nopa through local editorial and the NoPa neighborhood designation. The restaurant that named a neighborhood earns some discovery through that association even when platform coverage is thin.

§ 16 BIX 127 mentions

Bix

127

total mentions

A supper club that feels like 1940s Shanghai via North Beach. Claude dominates with 72 mentions — 57% of its total — and ChatGPT has 2. A restaurant that lives in the literary and cultural recommendation space rather than the algorithmic one. Whether a diner encounters Bix in an AI recommendation depends almost entirely on which app they open first.

ChatGPT2 mentions

2 mentions — the lowest ChatGPT count for any restaurant in the top 20. ChatGPT has essentially not discovered Bix in any of the June 2026 SF fine dining query contexts. A restaurant open since 1988 with no algorithmic footprint.

Claude72 mentions

Leads at 72 — 57% of Bix's entire dataset total. Claude is almost solely responsible for Bix's AI visibility, which for a supper club that has been a North Beach fixture since 1988 is a striking concentration. Claude recommends Bix; the others don't.

Gemini34 mentions

34 mentions. Gemini knows Bix and surfaces it in the right genre — cocktail bars, jazz venues, SF classics — with less enthusiasm than Claude but more presence than the other two platforms.

Perplexity19 mentions

19 — second-lowest. Bix's profile in Perplexity is limited. A restaurant that earns attention through atmosphere and cultural positioning rather than the editorial ranking patterns that drive Perplexity's discovery.

§ 17 KILN 116 mentions

Kiln

116

total mentions

Ron Siegel's wood-fire temple in the Presidio is earning serious attention — everywhere except Claude, which has given it exactly zero mentions in the June 2026 audit. ChatGPT leads at 54. That is an unusual reversal for a restaurant of this type: ChatGPT first, Claude missing entirely.

ChatGPT54 mentions

Leads at 54. Unusually for ChatGPT in this dataset, it is the platform that has discovered Kiln before the others — a reversal of the typical pattern where ChatGPT trails in editorial discovery. The Presidio location and the wood-fire concept appear to register in ChatGPT's recommendation set.

Claude0 mentions

Zero mentions across the entire June 2026 audit. Claude has not recommended Kiln once in the SF fine dining query set — the only restaurant in the top 20 with a platform at exactly zero. The Presidio location and the restaurant's relative newness may explain the gap.

Gemini38 mentions

38 mentions. Gemini knows Kiln and surfaces it in Presidio dining and wood-fire contexts, with ChatGPT well ahead. The platform discovers it in the right category but at lower frequency than ChatGPT — an unusual dynamic.

Perplexity24 mentions

24 mentions. Perplexity's Kiln coverage is limited — the restaurant may not yet have the editorial footprint that drives Perplexity's recommendation patterns. A newer restaurant without a dense critical record surfaces less in editorial-discovery contexts.

§ 18 SONS & DAUGHTERS 113 mentions

Sons & Daughters

113

total mentions

Quietly excellent for years in a tiny Tendernob dining room. ChatGPT, Gemini, and Perplexity are nearly even — 37, 37, 36. Claude has 3 mentions, making this the most dramatic Claude gap in the top 20 for a restaurant that clearly earns visibility everywhere else. Three platforms agree; one has essentially not found it.

ChatGPT37 mentions

37 and nearly tied with Gemini and Perplexity. ChatGPT, Gemini, and Perplexity are within one mention of each other — one of the most evenly distributed three-platform profiles in the top 20. The restaurant surfaces consistently everywhere Claude isn't looking.

Claude3 mentions

3 mentions against the others' combined 110. Claude has essentially not found Sons & Daughters — a Tendernob restaurant that receives consistent critical recognition. The gap is not accidental; Claude appears to miss a specific tier of mid-city fine dining that the other platforms surface.

Gemini37 mentions

37 and even with ChatGPT. Gemini surfaces Sons & Daughters in the fine dining tasting menu category, reliably present in a dataset where many restaurants earn nothing. The platform's coverage matches the restaurant's editorial positioning.

Perplexity36 mentions

36 — nearly even with ChatGPT and Gemini. Perplexity's balanced presence matches the editorial visibility of a restaurant with consistent local-press recognition. The three-platform agreement is the story here, not any individual platform's take.

§ 19 FOREIGN CINEMA 106 mentions

Foreign Cinema

106

total mentions

Irene Buenviaje's Mission institution has been projecting films and serving oysters since 1999. One of the most balanced profiles on this list — no platform leads by more than 14 mentions. Broadly findable, consistently itself. The most even platform distribution in the lower half of the top 20.

ChatGPT27 mentions

27 and even with Claude. ChatGPT's Foreign Cinema coverage emphasizes the outdoor patio, the film projections, and the Mission location — accurate and specific for a restaurant with a strong visual identity that translates into conventional descriptions.

Claude27 mentions

27 — even with ChatGPT. Claude describes Foreign Cinema in terms of atmosphere and the weekend brunch context — a slightly different emphasis from ChatGPT but the same count. The restaurant earns consistent coverage across platforms without leading any of them.

Gemini33 mentions

Leads at 33. Gemini's slight edge comes from describing the cinematic experience with more detail — the film projections, the open-air courtyard, the oyster bar setup. The platform that describes with more specificity leads by 6 mentions.

Perplexity19 mentions

19 — the lowest at 19. Foreign Cinema's profile in Perplexity is the most limited of the four, suggesting discovery happens through atmosphere and local knowledge rather than the editorial ranking patterns that drive Perplexity.

§ 20 EPIC STEAK 105 mentions

Epic Steak

105

total mentions

A waterfront steakhouse on the Embarcadero with a view that does most of the marketing. Gemini leads at 44 — more than double ChatGPT's count. Restaurants that photograph well tend to surface in visual-recommendation contexts first, and the Embarcadero setting gives Gemini something to describe that translates across query types.

ChatGPT20 mentions

20 mentions. ChatGPT's Epic Steak coverage is functional — the Embarcadero location, the waterfront view, the classic steakhouse format. Nothing wrong, nothing that couldn't apply to several other steakhouses in the city without revision.

Claude22 mentions

22 — near-even with ChatGPT. Claude mentions Epic Steak in special-occasion dining with a view contexts, which is the correct positioning for this restaurant. Similar count to ChatGPT, similar framing.

Gemini44 mentions

Leads at 44 — double ChatGPT's count. Gemini's Epic Steak coverage emphasizes the waterfront views and the visual setting, consistent with stronger coverage of visually distinctive environments. The bay views appear frequently in Gemini's descriptions.

Perplexity19 mentions

19 — the lowest at 19. Epic Steak's editorial profile is thinner relative to its visual distinctiveness, which may explain Perplexity's lower count. The restaurants Perplexity discovers tend to earn attention through critical writing rather than setting.