AidanBench
o1 3204.0 leads since December 2024
Not in model cards Models keep giving new answers to open-ended questions until answers become incoherent or repeat earlier ones. The count of fresh, coherent answers is the score, with no ceiling.
Rewards creativity and penalizes mode collapse. 322 GitHub stars. Prime Intellect hosts it as an environment.
- Made by
- Aidan McLaughlin (community)
- Released
- August 2024
- Kind
- Creative writing
- Level
- Everyday
- Scored
- Graded by a model
- Answers
- Open-ended
- Tasks
- All public
- Strongest model tested
- o1 #117 on the capability index
- Last model added
- February 2025
- Scores refreshed
- November 12th, 2026
25
Entries on its own leaderboard, 13 models we track
o1 leads with 3204.0 among models released by February 2025
Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.
The newest model with a score here came out in February 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- Claude Fable 5.2
-
Gemini 4 Carbon - GPT-6 Astra
- Claude Opus 5.5
- and 16 more
- 3204.0 o1
- 2691.0 claude-3.5-sonnet (snapshot unspecified)
- 1895.0 Gemini 2.0 Flash
- 1892.0 o1-preview
- 1577.0 o1-mini
- 1537.0 gpt-4-turbo (snapshot unspecified)
- 1531.0 Gemini 2.0 Flash Thinking
- 1515.0 gemini-pro-1.5 (snapshot unspecified)
- 1439.0 Claude 3.5 Sonnet
- 1333.0 GPT-4o
- 1331.0 gpt-4-1106-preview
- 1269.0 GPT-4o
- 1052.0 GPT-4o
- 1047.0 gemini-flash-1.5 (snapshot unspecified)
- 1028.0 gemma-2-27b-it
- 999.0 gpt-4-0314
- 953.0 mistral-large-latest (snapshot unspecified)
- 908.0 Claude 3 Opus
- 904.0 Claude 3.5 Haiku
- 900.0 chatgpt-4o-latest (snapshot unspecified)
- 859.0 gemma-2-9b-it
- 826.0 grok-beta
- 779.0
Llama 3.3 70B Instruct - 778.0 llama-3.1-405b-instruct
- 654.0 GPT-4o mini
- 1 o1 3204.0
- 2 claude-3.5-sonnet (snapshot unspecified) 2691.0
- 3 Gemini 2.0 Flash Feb 2025 1895.0
- 4 o1-preview 1892.0
- 5 o1-mini 1577.0
- 6 gpt-4-turbo (snapshot unspecified) 1537.0
- 7 Gemini 2.0 Flash Thinking Jan 2025 1531.0
- 8 gemini-pro-1.5 (snapshot unspecified) 1515.0
- 9 Claude 3.5 Sonnet 1439.0
- 10 GPT-4o Aug 2024 1333.0
- 11 gpt-4-1106-preview 1331.0
- 12 GPT-4o Nov 2024 1269.0
- 13 GPT-4o May 2024 1052.0
- 14 gemini-flash-1.5 (snapshot unspecified) 1047.0
- 15 gemma-2-27b-it 1028.0
- 16 gpt-4-0314 999.0
- 17 mistral-large-latest (snapshot unspecified) 953.0
- 18 Claude 3 Opus 908.0
- 19 Claude 3.5 Haiku 904.0
- 20 chatgpt-4o-latest (snapshot unspecified) 900.0
- 21 gemma-2-9b-it 859.0
- 22 grok-beta 826.0
- 23 Llama 3.3 70B Instruct 779.0
- 24 llama-3.1-405b-instruct 778.0
- 25 GPT-4o mini 654.0
Source: Official AidanBench leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | o1 | 3204.0 | Dec 2024 | Official leaderboard | |
| 2 | claude-3.5-sonnet (snapshot unspecified) | 2691.0 | Official leaderboard | ||
| 3 | Gemini 2.0 Flash Feb 2025 | 1895.0 | Feb 2025 | Official leaderboard | |
| 4 | o1-preview | 1892.0 | Sep 2024 | Official leaderboard | |
| 5 | o1-mini | 1577.0 | Sep 2024 | Official leaderboard | |
| 6 | gpt-4-turbo (snapshot unspecified) | 1537.0 | Official leaderboard | ||
| 7 | Gemini 2.0 Flash Thinking Jan 2025 | 1531.0 | Jan 2025 | Official leaderboard | |
| 8 | gemini-pro-1.5 (snapshot unspecified) | 1515.0 | Official leaderboard | ||
| 9 | Claude 3.5 Sonnet | 1439.0 | Jun 2024 | Official leaderboard | |
| 10 | GPT-4o Aug 2024 | 1333.0 | Aug 2024 | Official leaderboard | |
| 11 | gpt-4-1106-preview | 1331.0 | Official leaderboard | ||
| 12 | GPT-4o Nov 2024 | 1269.0 | Nov 2024 | Official leaderboard | |
| 13 | GPT-4o May 2024 | 1052.0 | May 2024 | Official leaderboard | |
| 14 | gemini-flash-1.5 (snapshot unspecified) | 1047.0 | Official leaderboard | ||
| 15 | gemma-2-27b-it | 1028.0 | Official leaderboard | ||
| 16 | gpt-4-0314 | 999.0 | Official leaderboard | ||
| 17 | mistral-large-latest (snapshot unspecified) | 953.0 | Official leaderboard | ||
| 18 | Claude 3 Opus | 908.0 | Feb 2024 | Official leaderboard | |
| 19 | Claude 3.5 Haiku | 904.0 | Oct 2024 | Official leaderboard | |
| 20 | chatgpt-4o-latest (snapshot unspecified) | 900.0 | Official leaderboard | ||
| 21 | gemma-2-9b-it | 859.0 | Official leaderboard | ||
| 22 | grok-beta | 826.0 | Official leaderboard | ||
| 23 | Llama 3.3 70B Instruct | 779.0 | Dec 2024 | Official leaderboard | |
| 24 | llama-3.1-405b-instruct | 778.0 | Official leaderboard | ||
| 25 | GPT-4o mini | 654.0 | Jul 2024 | Official leaderboard |
3 models have held number one on the official leaderboard since its release in August 2024
Lab claims count from the day the card came out, independent runs from the model’s release.
Anthropic OpenAI
Source: Official AidanBench leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
On the benchmark’s own leaderboard
Source: Official AidanBench leaderboard
No lab model card reports it yet
We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.
More in Writing and taste
- Design Bench (Moonshot) In use Models make graphic designs from direct prompts or few-shot reference images. An automated grader rates the results as Elo scores. Kimi K3.1 Pro 957
- WritingBench Fading Models complete 1,000 real-world writing requests across 6 domains and 100 subdomains. A judge model scores each reply on five query-specific criteria. GPT-5 83.9
- Creative Writing (MiniMax in-house) Only in system cards Models write creative pieces for real Hailuo AI user requests. Automatic assessments score the writing.
MiniMax-Text-01 81.3% - DeepSeek Chinese functional writing evaluation Only in system cards Models answer 3,170 everyday Chinese writing prompts such as reports, emails and speeches. Pairwise judging against a baseline model gives a win rate.
DeepSeek-V4-Pro 62.6% - Slide generation human evaluation (Z.ai) Only in system cards People compare HTML slide decks from two models on content, layout and visual quality. Metric is overall win rate. GLM-5 67.5%
- Task-dependent output homogenization evaluation Only in system cards Models answer prompts several times. A judge scores functional diversity where it is wanted and consistency where one answer is right. Muse Spark 0.21