Benchmark
FrontierScience Research
Scores are percentages. Higher is better.
- Models measured
- 8
- Scores recorded
- 9
- Sources
- 2
- IQ weight
- 2.20
Score distribution
- 26.7%
- Mean
- 28.3%
- Median
- 10.2%
- Std dev
- 21.3%
- 10th percentile
- 36.7%
- 90th percentile
- 3.3%
- Min
This benchmark has a weight of 2.20 for chat models, measured over 8 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 | Muse Spark· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 38.3% |
| 2 | GPT-5.4 ProOpenAI GPT-5.4 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 36.7% |
| 3 | Doubao Seed 2.1 Turbo· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 33.3% |
| 4 | GPT 5.4OpenAI GPT-5.4 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 33% |
| 5 | Doubao Seed 2.1 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 28.3% |
| 6 | GPT 5.2OpenAI GPT-5.4 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 25.2% |
| 7 | Hy3 (Deep Infra)· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 21.3% |
| 8 | Hy3 (Deep Infra)Hy3 official Benchmark Appendix · highest reasoning tier unless noted· Reported by Tencent Hunyuan· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 21.3% |
| 9 | Seed 2.0 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 3.3% |
Data credits: Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.