Benchmark
Draco
Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages. Higher is better.
- Models measured
- 14
- Scores recorded
- 17
- Sources
- 3
- IQ weight
- 0.81
Score distribution
- 74.5%
- Mean
- 77.2%
- Median
- 10.8%
- Std dev
- 59.8%
- 10th percentile
- 87%
- 90th percentile
- 46.7%
- Min
This benchmark has a weight of 0.81 for chat models, measured over 14 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 | Claude Opus 5· Steel.dev Agent Leaderboards data | 88.6% |
| 2 | Claude Opus 5.5· Steel.dev Agent Leaderboards data | 87.4% |
| 3 | Claude Sonnet 5.5· Steel.dev Agent Leaderboards data | 87% |
| 4 | ERNIE 5.5 Flash Preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 84.9% |
| 5 | Claude Haiku 5.5· Steel.dev Agent Leaderboards data | 81.5% |
| 6 | Claude Opus 4.8· Steel.dev Agent Leaderboards data | 80.4% |
| 7 | Claude Opus 4.7· Steel.dev Agent Leaderboards data | 77.7% |
| 8 | Claude Opus 4.7web search/fetch + code · adaptive thinking · max effort · 1M token limit · compaction at 200K· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 77.7% |
| 9 | Hy4 preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 77.2% |
| 10 | Hy4 previewofficial model-card appendix· Reported by Tencent Hy· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 77.2% |
| 11 | MiniMax M3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claimValues from 2 sources
| 73.2% |
| 12 | MiniMax M3MiniMax Code Deep Research Skill · official rubrics · Opus 4.6 grader· Reported by MiniMax Research· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 73.2% |
| 13 | Ling 3.0 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim | 70.4% |
| 14 | Claude Fable 5· Steel.dev Agent Leaderboards data | 65.3% |
| 15 | Claude Opus 4.6· Steel.dev Agent Leaderboards data | 59.8% |
| 16 | gemini-deep-research· Steel.dev Agent Leaderboards data | 59% |
| 17 | Claude Opus 4.5· Steel.dev Agent Leaderboards data | 46.7% |
Data credits: Steel.dev Agent Leaderboards, MIT (licence) · 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. · Data by LLM Stats