AI vs Puzzles

Back to Home
Puzzle#1720

Wordle

Released: 2026-03-05

7
Models Tested
6
Solved
4.71
Avg. Attempts
7
LLM Providers
Correct
Present
Absent

openai/gpt-4o

openai

3/6

c
r
a
n
e
e
l
o
p
e
s
h
e
e
p

mistralai/mistral-medium-3

mistral

Failed

c
r
a
n
e
d
e
p
t
h
h
e
l
p
s
s
p
e
n
d
s
p
e
c
k
s
p
e
l
l

google/gemini-2.0-flash-001

google

6/6

c
r
a
n
e
l
i
l
a
c
d
u
m
b
o
f
i
g
h
t
s
p
h
e
r
s
h
e
e
p

anthropic/claude-3.7-sonnet

anthropic

5/6

c
r
a
n
e
s
l
i
d
e
s
p
u
m
e
s
h
a
p
e
s
h
e
e
p

meta-llama/llama-3.3-70b-instruct

meta

3/6

h
o
u
s
e
h
a
s
t
e
s
h
e
e
p

deepseek/deepseek-chat-v3-0324

deepseek

6/6

c
r
a
n
e
s
l
o
p
e
s
w
e
e
p
s
l
e
e
p
s
t
e
e
p
s
h
e
e
p

x-ai/grok-3-mini-beta

x-ai

4/6

c
r
a
n
e
h
e
l
i
x
t
h
e
f
t
s
h
e
e
p

Methodology Note

Each model receives the same prompt structure with Wordle rules and feedback format. Models are given a maximum of 6 attempts with standard positional feedback after each guess to help refine their next attempt.

Leaderboard