AI vs Puzzles

Back to Home
Puzzle#1727

Wordle

Released: 2026-03-12

7
Models Tested
3
Solved
5.29
Avg. Attempts
7
LLM Providers
Correct
Present
Absent

mistralai/mistral-medium-3

mistral

Failed

c
r
a
n
e
d
e
p
t
h
b
e
g
e
t
l
e
v
e
e
e
l
i
d
e
l
e
e
c
h

openai/gpt-4o

openai

5/6

c
r
a
n
e
s
h
e
e
p
s
p
e
c
k
s
w
e
e
t
s
m
e
l
l

google/gemini-2.0-flash-001

google

Failed

r
a
i
s
e
l
u
t
e
s
s
h
e
w
s
s
l
i
p
s
s
l
e
e
k
s
o
l
e
s

anthropic/claude-3.7-sonnet

anthropic

5/6

s
l
a
t
e
s
p
i
r
e
s
h
o
w
n
s
m
e
l
t
s
m
e
l
l

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

meta

Failed

h
o
u
s
e
b
l
a
s
e
g
l
o
b
e
f
l
a
m
e
s
l
o
p
e
s
l
i
m
e

deepseek/deepseek-chat-v3-0324

deepseek

Failed

c
r
a
t
e
d
i
n
e
s
p
e
s
k
y
s
e
n
s
e
s
w
e
l
l
s
e
l
l
l

x-ai/grok-3-mini-beta

x-ai

3/6

c
r
a
n
e
m
o
t
e
l
s
m
e
l
l

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