Benchmark
ClockBench
Vision benchmarks measure how well models understand images. Scores are percentages.
- Models measured
- 30
- Scores recorded
- 30
- Sources
- 1
- IQ weight
- 2.76
Score distribution
- 26.3%
- Mean
- 15.6%
- Median
- 22.2%
- Std dev
- 7.8%
- 10th percentile
- 53.3%
- 90th percentile
- 3.3%
- Min
This benchmark has a weight of 2.76 for chat models, measured over 30 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 4 Carbonhigh· ClockBench | 85% |
| 2 | Gemini 4 Argonhigh· ClockBench | 83.3% |
| 3 | Gemini 3.8 Flash· ClockBench | 66.7% |
| 4 | Gemini 3.7 Flash· ClockBench | 53.3% |
| 5 | GPT 5.4high· ClockBench | 50.6% |
| 6 | GPT 5.5high· ClockBench | 46.1% |
| 7 | Muse Spark 1.3· ClockBench | 42.2% |
| 8 | Claude Fable 5· ClockBench | 37.2% |
| 9 | Gemini 3.5 Flash· ClockBench | 34.4% |
| 10 | Gemini 3.1 Pro Preview· ClockBench | 32.2% |
| 11 | Gemini 3 Pro Preview· ClockBench | 31.1% |
| 12 | Grok 4.5· ClockBench | 21.7% |
| 13 | Grok 4.6· ClockBench | 20% |
| 14 | Gemini 2.5 Pro· ClockBench | 18.9% |
| 15 | GPT 5.2high· ClockBench | 15.6% |
| 16 | Claude Opus 4.7· ClockBench | 15% |
| 17 | O3 Pro· ClockBench | 14.4% |
| 18 | Qwen3 VL 235Bthinking· ClockBench | 14.4% |
| 19 | OpenAI o3high· ClockBench | 12.2% |
| 20 | Gemini 2.5 Flash· ClockBench | 11.1% |
| 21 | GPT 5high· ClockBench | 11.1% |
| 22 | GPT-5 Pro· ClockBench | 11.1% |
| 23 | Mistral Medium 3.1· ClockBench | 10.6% |
| 24 | Claude Opus 4.6· ClockBench | 10% |
| 25 | Claude Opus 4.1· ClockBench | 8.9% |
| 26 | GPT-5 mini· ClockBench | 8.9% |
| 27 | Claude Sonnet 4.5· ClockBench | 7.8% |
| 28 | Claude Sonnet 4· ClockBench | 6.7% |
| 29 | Qwen2.5 VL 72B Instruct· ClockBench | 6.1% |
| 30 | Grok 4 Fast· ClockBench | 3.3% |