Benchmark
RISEBench - Spatial
Scores are percentages.
- Models measured
- 12
- Scores recorded
- 14
- Sources
- 1
- IQ weight
- 0.68
Score distribution
- 28.5%
- Mean
- 31%
- Median
- 16.5%
- Std dev
- 11%
- 10th percentile
- 50%
- 90th percentile
- 2%
- Min
This benchmark has a weight of 0.68 for image-generation models, measured over 10 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 Image 1.5· RISEBench data | 62% |
| 2 | GPT Image 2· RISEBench data | 50% |
| 3 | Gemini 3 Pro Image· RISEBench data | 48% |
| 4 | Gemini 2.5 Flash Image· RISEBench data | 37% |
| 5 | GPT Image 1· RISEBench data | 37% |
| 6 | GPT-Image-1 mini· RISEBench data | 33% |
| 7 | Qwen Image Edit2511· RISEBench data | 31% |
| 8 | Gemini 2.0 Flash Experimental· RISEBench data | 23% |
| 9 | BAGELw/ CoT· RISEBench data | 21% |
| 10 | Qwen Image Edit2509· RISEBench data | 17% |
| 11 | BAGEL· RISEBench data | 14% |
| 12 | Flux Kontext Dev· RISEBench data | 13% |
| 13 | Doubao Seedream 4-0· RISEBench data | 11% |
| 14 | Step1X-Edit· RISEBench data | 2% |