THOR Finding Triage - Critical Miss Rate: leaderboard
Nextron's THOR Finding Triage Benchmark asks models to triage 189 expert-labelled findings from THOR forensic scanner reports. Critical Miss Rate is the percentage of true-positive findings the model suppresses as false positives; lower is better.
Metric: True positives suppressed as false positives (%, lower is better). Source: nextron-labs.github.io. Saturation forecast: Estimated already saturated. 89 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 0 | #159 |
| 2 | GPT-5 | 0 | #105 |
| 3 | Gemini 2.5 Flash | 0 | #255 |
| 4 | Claude Sonnet 4.6 | 0 | #94 |
| 5 | GPT-5.4 | 0 | #83 |
| 6 | GPT-5.5 | 0 | #33 |
| 7 | GPT-4o Mini | 0 | #598 |
| 8 | Claude Opus 4.6 | 0 | #67 |
| 9 | DeepSeek V4 Pro | 0 | #112 |
| 10 | GPT-5 Nano | 0 | #439 |
| 11 | GPT-5 Mini | 0 | #193 |
| 12 | Qwen 3 32B | 0 | #449 |
| 13 | Claude Opus 4.8 | 0 | #39 |
| 14 | Qwen 3 14B | 0 | #542 |
| 15 | Gemma 3 27B | 0 | #606 |
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-critical-miss-rate · How It Works · Data refreshed daily, snapshot 2026-11-05.