Released: 2026-02-18
mistralai/mistral-medium-3
mistral
2/4
CRIMP, CURL, TEASE, SHAVING
HAIR STYLING TERMS
BAD, FLY, RAD, WICKED
SLANG TERMS
BANTAM, CRESTED, FREE-RANGE, LEGHORN
TYPES OF CHICKENS
FEATHER, HEAVY, SOUR, TOPICAL
TYPES OF CREAM
google/gemini-2.0-flash-001
2/4
BAD, HEAVY, RAD, WICKED
Slang terms for ’good’
BANTAM, CRESTED, FREE-RANGE, LEGHORN
Types of Chickens
CURL, CRIMP, FEATHER, TEASE
Hair Styling Actions
FLY, SHAVING, SOUR, TOPICAL
Types of Foam
openai/gpt-4o
openai
2/4
CRIMP, CURL, FEATHER, TEASE
Hairstyle techniques
BAD, RAD, WICKED, HEAVY
Slang for good or cool
BANTAM, CRESTED, FREE-RANGE, LEGHORN
Types of chickens
FLY, SHAVING, SOUR, TOPICAL
Adjectives describing various subjects
anthropic/claude-3.7-sonnet
anthropic
4/4
CRIMP, CURL, FEATHER, TEASE
HAIR STYLING TECHNIQUES
BAD, FLY, RAD, WICKED
SLANG FOR COOL/AWESOME
BANTAM, CRESTED, FREE-RANGE, LEGHORN
TYPES OF CHICKENS
HEAVY, SHAVING, SOUR, TOPICAL
TYPES OF CREAM
meta-llama/llama-3.3-70b-instruct
meta
1/4
CRIMP, CURL, TEASE, SHAVING
Hair styles or related terms
BAD, RAD, WICKED, HEAVY
Informal terms for something good or excellent
BANTAM, CRESTED, LEGHORN, FREE-RANGE
Types of chickens
FEATHER, FLY, SOUR, TOPICAL
Types of cuts or styles in food or other contexts, with FLY potentially referring to a type of fishing lure and SOUR relating to a type of cream or other food product, and TOPICAL referring to a type of application or treatment
deepseek/deepseek-chat-v3-0324
deepseek
4/4
BAD, FLY, RAD, WICKED
Slang terms meaning ’cool’ or ’excellent’
BANTAM, CRESTED, FREE-RANGE, LEGHORN
Types of chickens
CRIMP, CURL, FEATHER, TEASE
Hair styling techniques
HEAVY, SHAVING, SOUR, TOPICAL
Types of cream (heavy cream, shaving cream, sour cream, topical cream)
x-ai/grok-3-mini-beta
x-ai
4/4
CRIMP, CURL, TEASE, FEATHER
Hair Styling
BAD, RAD, WICKED, FLY
Slang for Cool
BANTAM, CRESTED, FREE-RANGE, LEGHORN
Chicken Types
HEAVY, SHAVING, SOUR, TOPICAL
Cream Types
Each model receives the same prompt with the 16 words to connect. Models are tasked with creating 4 connections and their answers are evaluated without feedback or retries.