Benchmark
Longform Creative Writing
Scores are percentages.
Scores from EQ-Bench boards.
- Models measured
- 124
- Scores recorded
- 124
- Sources
- 1
- IQ weight
- 1.57
Score distribution
- 59.2%
- Mean
- 60.2%
- Median
- 17.4%
- Std dev
- 32.9%
- 10th percentile
- 81.5%
- 90th percentile
- 17%
- Min
This benchmark has a weight of 1.57 for chat models, measured over 124 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 101 | Reka Flash 3· EQ-Bench boards | 41.2% |
| 102 | Qwen3 32B· EQ-Bench boards | 40.5% |
| 103 | Grok 4.1 Fast· EQ-Bench boards | 40.1% |
| 104 | Nova 2 Litev1· EQ-Bench boards | 38.8% |
| 105 | QwQ 32B· EQ-Bench boards | 36.9% |
| 106 | Qwen3 14B· EQ-Bench boards | 35.9% |
| 107 | GPT OSS 120B· EQ-Bench boards | 35.8% |
| 108 | Qwen3 30B A3B· EQ-Bench boards | 35.6% |
| 109 | Trinity Large Preview· EQ-Bench boards | 35% |
| 110 | MiniMax-M2 Her· EQ-Bench boards | 34.8% |
| 111 | Gemma 3 4B IT· EQ-Bench boards | 34.4% |
| 112 | Nemotron 3 Nano 30B A3B· EQ-Bench boards | 32.9% |
| 113 | Mistral Small Creative· EQ-Bench boards | 32.6% |
| 114 | GPT-4.1 nano· EQ-Bench boards | 32.4% |
| 115 | Llama 4 Maverick· EQ-Bench boards | 31.9% |
| 116 | Ministral 3 14B Instruct· EQ-Bench boards | 31.9% |
| 117 | Qwen3 Next 80B A3B Instruct· EQ-Bench boards | 31.5% |
| 118 | Llama 4 Scout· EQ-Bench boards | 31.1% |
| 119 | Mistral Nemo· EQ-Bench boards | 28.7% |
| 120 | GPT-5 nano· EQ-Bench boards | 28% |
| 121 | Solar Pro 3· EQ-Bench boards | 27% |
| 122 | Phi 4 Multimodal Instruct· EQ-Bench boards | 22.8% |
| 123 | GPT OSS 20b· EQ-Bench boards | 21.2% |
| 124 | Qwen3 4B· EQ-Bench boards | 17% |
Showing 101–124 of 124 results.
Data credits: EQ-Bench (Samuel J. Paech) — cite paech2023eqbench