Benchmark
RISEBench - Instruction Reasoning
Scores are percentages.
- Models measured
- 12
- Scores recorded
- 14
- Sources
- 1
- IQ weight
- 0.91
Score distribution
- 52%
- Mean
- 54.1%
- Median
- 16%
- Std dev
- 26%
- 10th percentile
- 73.8%
- 90th percentile
- 25.1%
- Min
This benchmark has a weight of 0.91 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 | Gemini 3 Pro Image· RISEBench data | 77% |
| 2 | GPT Image 2· RISEBench data | 73.8% |
| 3 | GPT Image 1.5· RISEBench data | 69.7% |
| 4 | GPT Image 1· RISEBench data | 62.8% |
| 5 | Gemini 2.5 Flash Image· RISEBench data | 61.2% |
| 6 | Doubao Seedream 4-0· RISEBench data | 58.9% |
| 7 | GPT-Image-1 mini· RISEBench data | 54.1% |
| 8 | Qwen Image Edit2511· RISEBench data | 49.9% |
| 9 | Gemini 2.0 Flash Experimental· RISEBench data | 48.9% |
| 10 | BAGELw/ CoT· RISEBench data | 47.6% |
| 11 | Qwen Image Edit2509· RISEBench data | 37.2% |
| 12 | BAGEL· RISEBench data | 36.4% |
| 13 | Flux Kontext Dev· RISEBench data | 26% |
| 14 | Step1X-Edit· RISEBench data | 25.1% |