AI vs Puzzles

Back to Home
Puzzle#224

Geoguessr

Released: 2026-06-16

Google Street View Image
Country Flag

Namibia

Lat: -19.11207081374207

Lng: 16.13290570929091

See on Google Street View


6
Models Tested
492.2km
Avg. Distance
3958.5
Avg. Score
5
LLM Providers

Key:

Attempt
Solution

Score Threshold:

>4400
>2000
<2000

anthropic/claude-haiku-4.5

anthropic

Score: 3575

Distance: 671km

Lat: -22.3284

Lng: 21.5936


This image shows a typical unpaved dirt road in an arid/semi-arid region with sparse vegetation. The landscape features scattered acacia-like trees, dry grassland, and reddish soil characteristic of southern Africa. The vegetation pattern, soil color, road style, and overall landscape appearance are consistent with the Kalahari region, specifically Botswana. The wind-swept trees and desert vegetation suggest this is likely along one of the main routes in central or southern Botswana, possibly in or near the Central Kalahari Game Reserve area or along roads in that region.


meta-llama/llama-4-maverick

meta

Score: 3801

Distance: 548km

Lat: -23.593306

Lng: 18.343506


The presence of the ’Chris du Plessis’ watermark in the bottom-right corner of the image is a strong indicator that the image is located in Namibia, as Chris du Plessis is known to be a Street View contributor who has captured imagery in Namibia. The arid landscape and dirt road are consistent with the geography of Namibia. By searching for similar images and coordinates associated with Chris du Plessis’ contributions, the location can be narrowed down to the Kalahari Desert region in southern Namibia.


anthropic/claude-sonnet-4.6

anthropic

Score: 3964

Distance: 464km

Lat: -19.8

Lng: 20.5


The image shows a dusty dirt road in an arid savanna landscape with sparse acacia-type trees, sandy/reddish soil, and clear blue sky. The photographer credit ’Chris du Plessis’ suggests a southern African location. The vegetation type (sparse bushveld/savanna), the reddish sandy soil, and the overall landscape appearance strongly suggest Botswana or Namibia. The Kalahari or Etosha region fits well with this type of terrain. Given the contributor name and landscape characteristics, this appears to be in Botswana, likely in the Kalahari region or near Etosha area of Namibia.


google/gemini-3-flash-preview

google

Score: 4851

Distance: 60km

Lat: -18.784262

Lng: 16.591244


The image displays a dirt road with light-colored, alkaline-rich soil characteristic of the Etosha Pan region. The vegetation consists of mopane trees (Colophospermum mopane) and scrub typical of northern Namibia. The photographer credit to Chris du Plessis is strongly associated with extensive Street View coverage of South African and Namibian national parks, specifically Etosha National Park.


openai/gpt-5-image-mini

openai

Score: 2982

Distance: 1034km

Lat: -27.5

Lng: 20.5


Arid, flat scrubland with pale sandy/red road and low thorny trees (acacia-like) typical of the Kalahari thornveld. Image credit visible for Chris du Plessis (South African Google imagery contributor), pointing to southern Africa. Vegetation, soil color and openness match Northern Cape/Kgalagadi region of South Africa/Namibia border, so I place the guess in the Kalahari near Upington/Kgalagadi (approx. -27.5, 20.5).


MoonshotAI

moonshotai/kimi-k2.5

moonshot

Score: 4578

Distance: 176km

Lat: -18.2

Lng: 17.5


The image shows a flat, semi-arid landscape with a dirt road and scattered acacia/mopane trees characteristic of the African savanna. The copyright attribution to ’Chris du Plessis’ strongly suggests Namibia, as this person is known for contributing extensive Street View imagery in Namibia, particularly in the northern regions. The flat terrain, white/gray soil, and vegetation density are typical of northern Namibia (Owambo or Kunene regions), which features extensive plains with scattered trees and unpaved roads. The clear blue sky and dry conditions align with Namibia’s climate. The landscape is less hilly than South Africa’s comparable regions and matches the Owambo plains area near the Angolan border.


Methodology Note

Each model receives the same Google Street View image. Models are tasked with guessing the location of the image using only visual clues from the street view. Their answers are evaluated without feedback or retries.

Leaderboard