Benchmark
SMDD-Bench
Scores are percentages.
Scores from Harbor Hub Leaderboards.
- Models measured
- 7
- Scores recorded
- 7
- Sources
- 1
- IQ weight
- 0.76
Score distribution
- 31.2%
- Mean
- 30.3%
- Median
- 7.5%
- Std dev
- 24.3%
- 10th percentile
- 39%
- 90th percentile
- 19.3%
- Min
This benchmark has a weight of 0.76 for chat models, measured over 7 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 |
|---|---|---|
| 1 | GPT 5.4main· Harbor Hub Leaderboards | 40.2% |
| 2 | Gemini 3.1 Pro Previewmain· Harbor Hub Leaderboards | 39% |
| 3 | Claude Sonnet 4.6main· Harbor Hub Leaderboards | 38% |
| 4 | Kimi K2.5main, thinking· Harbor Hub Leaderboards | 30.3% |
| 5 | Qwen3.5 397B A17Bmain· Harbor Hub Leaderboards | 27.5% |
| 6 | DeepSeek V3.2main· Harbor Hub Leaderboards | 24.3% |
| 7 | MiniMax M2.7main· Harbor Hub Leaderboards | 19.3% |