THOR Finding Triage - CW%: leaderboard
Nextron's THOR Finding Triage Benchmark asks models to triage 189 expert-labelled findings from THOR forensic scanner reports. CW% is the confidence-weighted classification score in percent; higher is better. The authors advise reading it with Balanced OTS and Critical Miss Rate; naive always-one-class baselines are excluded.
Metric: Confidence-weighted classification score (%). Source: nextron-labs.github.io. Saturation forecast: Around November 2027. 89 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.7 Flash | 72.5 | #24 |
| 2 | Gemini 3.5 Flash | 69.6 | #66 |
| 3 | Gemini 3.6 Flash | 68.7 | #57 |
| 4 | Gemini 3.1 Flash Lite | 67.8 | #200 |
| 5 | Gemini 3.8 Flash | 67.4 | #30 |
| 6 | Claude Opus 4.5 | 66.3 | #87 |
| 7 | Gemini 3.1 Pro (Preview) | 66.2 | #64 |
| 8 | Qwen 3.6 27B | 66.1 | #182 |
| 9 | Gemma 4 31B | 65.7 | #212 |
| 10 | Qwen 3.7 Max | 65.7 | #85 |
| 11 | Qwen 3.6 Max | 65.4 | #97 |
| 12 | GLM-5.1 | 65.3 | #109 |
| 13 | DeepSeek V4.1 Flash (1017) | 64.7 | #27 |
| 14 | GLM-5 | 64.2 | #149 |
| 15 | Kimi K2.7 Code | 64 | #95 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=thor-finding-triage-cw · How It Works · Data refreshed daily, snapshot 2026-11-05.