▤
Pentest report → exec summary
SecurityTurn a dense technical pentest report into a one-page summary a CISO or a customer can read — posture, top risks, and what’s being fixed.
The routing recipe
Job typeSummarizationPolicyBalanced
routes to
NVIDIA Nemotron 3 Super 120B-A12B nvidia
nvidia:nvidia/nemotron-3-super-120b-a12b
86.3
match score
84.8
task fit
100
cost
70
speed
Why this model from benchmarks & capabilities
- "Summarization" leans hardest on instruction; NVIDIA Nemotron 3 Super 120B-A12B scores 85/100 there — #14 of 40.
- Strength on this class of work shows up on LiveBench (language), grounding & summarization suite, where NVIDIA models are competitive.
- Holds the 16k+ context this job needs (this model: 262k).
- Cost: runs on NVIDIA free credits / self-hostable NIM — effectively $0 marginal. Speed: ~150 tok/s (estimated).
- Chosen over GPT-OSS 20B for higher task-fit (84.8 vs 75.2) & more context.
| Capability the job needs | Weight | NVIDIA Nemotron 3 Super 120B-A12B | Catalog rank |
|---|---|---|---|
| instruction | 0.90 | #14 of 40 | |
| long_context | 0.90 | #1 of 40 | |
| speed | 0.50 | #15 of 40 |
Relevant benchmarksLiveBench (language) grounding & summarization suite
Why it qualifiesHolds the 16k+ context this job needs (this model: 262k).
EconomicsCost: runs on NVIDIA free credits / self-hostable NIM — effectively $0 marginal.
Speed: ~150 tok/s (estimated). vs runner-upChosen over GPT-OSS 20B for higher task-fit (84.8 vs 75.2) & more context. (GPT-OSS 20B, score 85.7)
Speed: ~150 tok/s (estimated). vs runner-upChosen over GPT-OSS 20B for higher task-fit (84.8 vs 75.2) & more context. (GPT-OSS 20B, score 85.7)
Example result
Sample input ▾
Report: 1 High (reflected XSS on /search, remediated + retested), 3 Medium (missing security headers, verbose error messages, outdated jQuery), 5 Low/Info.
Posture: Strong. No critical issues; the single High finding was fixed and re-tested before the report closed.
By severity: 1 High (resolved), 3 Medium, 5 Low/Info.
Top findings: (1) a cross-site scripting bug on search — fixed; (2) missing browser security headers — quick config change; (3) an outdated front-end library — scheduled upgrade.
Remediated: the High and one Medium. Remaining 2 Mediums are config/library updates targeted within 30 days; Lows within 90. Full report and retest letter available under NDA.
Test it on your own data