1
openai/gpt-5.3-codex
3.6000
WeirdBench
Benchmark leaderboard
wordle
Wordle
Play 20 recent Wordle answers turn by turn with standard gray/yellow/green feedback. Invalid guesses still cost a turn, scores are capped at 10 turns per puzzle, and lower is better.
Lower score is better
2
openai/gpt-5.3-chat
3.8000
3
anthropic/claude-opus-4.6
4.0500
4
openai/gpt-oss-120b
4.2000
5
inception/mercury-2
4.5500
6
anthropic/claude-sonnet-4.5
4.6500
7
anthropic/claude-opus-4.5
5.5500
8
google/gemini-3-flash-preview
6.8000
9
anthropic/claude-opus-4.1
6.8500
10
google/gemini-3.1-flash-lite-preview
8.2000
11
anthropic/claude-haiku-4.5
9.2000
12
openai/gpt-5.4
9.3500
13
x-ai/grok-4.20-beta
9.6000
14
deepseek/deepseek-v3.2
9.7000
15
openai/gpt-5.4-mini
9.7000
16
mistralai/mistral-medium-3.1
9.7500
17
openai/gpt-5.1
9.7500
18
mistralai/mistral-large-2512
9.8500
19
meta-llama/llama-4-maverick
9.9500
20
amazon/nova-2-lite-v1
10.0000
21
amazon/nova-lite-v1
10.0000
22
amazon/nova-micro-v1
10.0000
23
amazon/nova-pro-v1
10.0000
24
meta-llama/llama-4-scout
10.0000
25
mistralai/mistral-small-2603
10.0000
Methodology
How scoring works
Use a fixed set of 20 recent Wordle answers, run a fresh chat loop for each puzzle, and score the average turns needed to solve while applying standard duplicate-letter feedback rules.
Prompt
The model is told to reply with exactly one 5-letter word per turn, that any extra text is penalized, and that duplicate letters are allowed.
Score
Lower is better. Each puzzle score is the turn the word is solved on, or 10 if the model never solves it within 10 turns. Invalid guesses still count as turns.
Execution
Benchmark runners execute locally, simulate the Wordle judge deterministically, cache results in Neon, and skip recomputation for models that already have stored scores.